Trees, Fungi, Insects: How Host Plant Genetics Builds a Community

Graduate Theses, Dissertations, and Problem Reports

2020

Sandra Jeanne Simon West Virginia University, [email protected]

Follow this and additional works at: https://researchrepository.wvu.edu/etd

Part of the Entomology Commons, Evolution Commons, Genetics Commons, Plant Biology Commons, and the Population Biology Commons

Recommended Citation Simon, Sandra Jeanne, "Trees, Fungi, Insects: How Host Plant Genetics Builds a Community" (2020). Graduate Theses, Dissertations, and Problem Reports. 7779. https://researchrepository.wvu.edu/etd/7779

This Dissertation is protected by copyright and/or related rights. It has been brought to you by the The Research Repository @ WVU with permission from the rights-holder(s). You are free to use this Dissertation in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you must obtain permission from the rights-holder(s) directly, unless additional rights are indicated by a Creative Commons license in the record and/ or on the work itself. This Dissertation has been accepted for inclusion in WVU Graduate Theses, Dissertations, and Problem Reports collection by an authorized administrator of The Research Repository @ WVU. For more information, please contact [email protected]. Graduate Theses, Dissertations, and Problem Reports

2020

Trees, Fungi, Insects: How Host Plant Genetics Builds a Community

Sandra Jeanne Simon

Follow this and additional works at: https://researchrepository.wvu.edu/etd

Part of the Entomology Commons, Evolution Commons, Genetics Commons, Plant Biology Commons, and the Population Biology Commons Trees, Fungi, Insects: How Host Plant Genetics Builds a Community

Sandra Jeanne Simon

Dissertation submitted to the Eberly College of Arts and Sciences at West Virginia University in partial fulfillment of the requirements for the degree of

Doctor of Philosophy In Biology

Stephen P. DiFazio, Ph.D., Chair Yong-Lak Park, Ph.D. Gina M. Wimp, Ph.D. Jonathan R. Cumming, Ph.D. Jennifer S. Hawkins, Ph.D.

Morgantown, West Virginia 2020

Keywords: Populus, Salix, biotic interactions, comparative genomics, tandem duplication, community genetics, floral visitors, Andrena, phenolics, volatile organic compounds

Trees, Fungi, Insects: How Host Plant Genetics Builds a Community

Sandra Jeanne Simon

Organisms, such as fungi and insects, can cause millions of acres of agricultural and forest damage, while others provide billions of dollars in ecological services such as education, aesthetic enjoyment, pollination, and gardening. Plant breeding and biotechnology can potentially help establish a balance between the proliferation of detrimental pests and attraction of beneficial insects. Variation in plant physiological and morphological characteristics are extremely important in the ability of host tissues to support many different types of organisms. When that variation is genetically heritable in a plant population, shifts in the underlying genes can have predictable consequences in structuring entire ecosystems. The field of community genetics seeks to study these interactions and identify the genes important in host plants, which will ultimately allow for the prediction of community level responses to changing conditions. The main goal of my dissertation was to identify the genetic underpinnings of host plant-biotic community organization in species belonging to the Salicaceae family, which contains many species of trees and shrubs of ecological and economic importance. To date, community genetic research has established the ability of hybrid plants to have wide-ranging heritable effects on communities and ecosystems. However, only a few publications have identified the genes underlying these relationships in pure species.

In chapter 2, I utilized a pseudo-backcross hybrid family of Populus and quantitative trait analysis (QTL) as well as genomic comparisons of the P. trichocarpa and P. deltoides parents to identify potential candidate genes mediating their relationship with several insect herbivores and fungal pathogens. I found that many gene candidates had undergone recent tandem duplication and this pattern was enriched relative to the rest of the genome in the native parent QTL intervals. Additionally, I found the hybrids were mediating interactions between pathogens leading to unique genetic associations that would not normally be observed in a single species population, which may contribute to the elevated community effects that have been previously observed in natural hybrid zones. In chapter 3, I used surveys that were conducted in multiple common gardens of a population of P. trichocarpa, genome wide association analysis (GWAS), and networks to identify genes and potential biological functions underlying arthropod community composition. I found that genes associated with individual arthropods appeared to be very functionally targeted with rare variants related to metabolite production and manipulation of tissue nutrition. Genes that associated with arthropod richness and community composition have biological functions that may allow them to more broadly target multiple groups of arthropods, such as terpenoid synthesis, RNA inhibition, and transmembrane protein activity. In chapter 4, I used visual observation and pan-traps to survey the tree species Salix nigra and explore the impact of dioecy on the assembly of floral insect communities. I found that male trees supported higher diversity of floral visitors on their catkins when compared to females due to visual cues of yellow pollen. I also identified the main cross-pollinators to be three species of Andrena bees, one of which (A. nigrae) showed a preference for female flowers and was correlated to specific VOC cues from catkins. Finally, I detected an asynchrony in catkin bloom and insect emergence in early spring that threatens not only the sexual reproduction of S. nigra trees, but also the

survival of local A. nigrae populations. Overall, I found that the dynamic plant-pathogen- herbivore-pollinator relationships are dependent on combinations of plant genetic effects with spatial and temporal environmental variability.

Acknowledgements

Thank you, Steve, for all your guidance on my project and patience in helping me write this dissertation. I would never have stayed at WVU if not for your support and confidence that I would be successful in science, and I wouldn’t have been successful without you ensuring I was always well hydrated in the field and well caffeinated in the office. I also would like to thank my committee members for all your guidance and input. In particular I would like to thank Dr. Wimp for all the phone time you carved out of your schedule, whether it be normal working hours or spring break, to drop everything and talk about statistics, bugs, trees and how to find bugs on trees; Dr. Park for always being available to critique my methods and help me to improve them to capture the best version of an insect community possible for my research (and always reminding me that larvae and nymphs are not the same!); Dr. Cumming for showing me during my undergrad a look at how the experimental process develops (with an equal number of failures and successes) and sharing your knowledge of plant stress responses which helped me to structure my research; and finally Dr. Hawkins for supporting me though the Genomics Core assistantships in my final years, having one of my favorite labs in the department to collaborate with, and reminding me to fill out my defense declaration form which I am doing as I type out these acknowledgements.

I would like to thank my lab including Ran Zhou, Hari Chhetri, David Macaya-Sanz, Roshan Abeyratne, and Julianne Grady. I can’t imagine a better group of people to work with in science than all of you. Also, thanks to Ryan Percifield, I learned so much about laboratory technique from you during my time in the Genomics Core it wouldn’t be a stretch to say it is one of my most valuable skill sets and an easy one to have sacrificed a lot of chocolate to acquire. Additionally, for all the people that helped me with the field and lab surveys including Dr. Jared LeBoldus, Dr. Luke Evans, Dr. Ken Keefover-Ring, Dr. William MacDonald, Dr. Nesatalu Hiese, Jacob Miller, Margo Folwick, Tanita Cheevaphantusri, and Sunita Mahat.

I want to give a special shout-out to two people that I could talk to for hours about any kind of science. Ashley Henderson you have been a constant joy to work with and an even better friend through the hot mess express that is a doctorate in biology. Brandon Sinn you are easily one of the most unique and talented researchers I have had the opportunity to work closely with, I hope your career continues to inspire you the same way you did me in my last few semesters. Finally, I want to acknowledge Mark Burnham (and Maya) who has been a constant support through the end of my degree. You have helped me turn countless jumbled thoughts into many of the coherent ideas conveyed throughout this document. After my defense, I promise I will wait at least one hour after waking up before I start asking you science questions in the morning. Also, to all my co-authors for the countless hours they have spent helping collect data and edit my research chapters for final publication. Funding for this work came from the Bioenergy Science Corporation (BESC), the Center for Bioenergy Innovation (CBI), the National Science Foundation (NSF) Dimensions of Biodiversity, the Morrissey Ropp Scholarship, the Core Research in Taxonomy of Vascular Plants Scholarship, the Department of Biology WVU genomics Core graduate fellowship, and the WVU Department of Biology and Eberly College of Arts and Sciences.

iv Table of Contents

Abstract ...... ii Acknowledgements ...... iv Table of Contents ...... v List of Tables and Figures ...... vii Chapter 1: Introduction: Biotic interactions in the Salicaceae ...... 1 1.1 Community genetics ...... 2 1.2 The Salicaceae family as a genetic model ...... 3 1.3 Physiological and morphological features ...... 4 1.4 Dioecy and floral visitor communities ...... 5 1.5 Objectives ...... 7 1.6 Literature cited ...... 9 Chapter 2: Host plant genetic control of associated fungal and insect species in a Populus hybrid ...... 14 2.1 Abstract ...... 15 2.2 Introduction ...... 16 2.3 Methods ...... 18 2.4 Statistical analysis ...... 21 2.5 Results ...... 25 2.6 Discussion ...... 28 2.7 Literature cited ...... 38 2.8 Tables and Figures ...... 48 Chapter 3: Genetic underpinnings of arthropod community distributions in Populus trichocarpa ...... 65 3.1 Abstract ...... 66 3.2 Introduction ...... 67 3.3 Methods ...... 69 3.4 Statistical analysis ...... 70 3.5 Results ...... 73 3.6 Discussion ...... 75 3.7 Literature cited ...... 81 3.8 Tables and Figures ...... 91 v Chapter 4: Characterization of Salix nigra floral insect community and activity of three native Andrena bees ...... 111 4.1 Abstract ...... 112 4.2 Introduction ...... 113 4.3 Methods ...... 115 4.4 Statistical analysis ...... 119 4.5 Results ...... 123 4.6 Discussion ...... 126 4.7 Literature cited ...... 132 4.8 Tables and Figures ...... 136 Chapter 5: Conclusions ...... 160 5.1 Overview of research goals ...... 161 5.2 Hybrid mediation of biotic associations ...... 161 5.3 P. trichocarpa community genetics ...... 163 5.4 S. nigra floral community ...... 164 5.5 Overall conclusions ...... 165 5.6 Literature cited ...... 167

vi List of Tables and Figures

Table 2-1. TPS and linear-mixed model output for biotic surveys. Broad-sense heritability (H2) denotes the contribution of all host plant genetic factors to total variance in the biotic phenotype. R package RLRsim exactRLRT function was used to test significance of effects in mixed model...... 48 Table 2-2. Summary of QTL permutation test output. Percent variance in surveys for insects and fungi explained by significant marker indicated for composite interval mapping models. Positive (+) allele specifies genotype at significant interval that results in an increase in susceptibility. D indicates progeny are homozygous for P. deltoides alleles and T indicates progeny are heterozygous for P. deltoides and P. trichocarpa alleles...... 49 Table 2-3. One-way ANOVA, with Genotype as a covariate, analyzing the effect of infection severity of the genus of one fungus on the infection severity of the competing leaf fungi in 2008...... 50 Table 2-4. Number of genes in QTL intervals in parental genomes for biotic associations...... 51

Table 2-5. Candidate genes under positive selection (Ka/Ks>1) in genetic intervals associated with fungi and insects...... 52 Table 2-6. Tandem duplication profiles for genetic intervals. Number of copies next to species gene name indicates the size of tandem expansion for the gene...... 53 Table 3-1. Sum of arthropods observed on tree tissue within plantations...... 91 Table 3-2. Linear-mixed model output for community surveys across all gardens and years. Broad-sense heritability (H2) denotes the contribution of all host plant genetic factors to total variance in phenotype...... 92 Table 3-3. PERMANOVA results for NMDS configuration within all 2012 gardens with replicated genotype observations and block to account for spatial position...... 93 Table 3-4. PERMANOVA results for NMDS configuration among gardens with replicated genotype observations and block to account for spatial position...... 94 Table 3-5. Significant SNPs (FDR < 0.1) and nearest P. trichocarpa genes discovered through GWAS for P. populiella trait in Clatskanie garden...... 95 Table 3-6. Significant SNPs (FDR < 0.1) and nearest P. trichocarpa genes discovered through GWAS for Phyllonorycter sp. trait...... 96 Table 3-7. Significant SNPs (FDR < 0.1) and closest P. trichocarpa genes discovered through GWAS for Community metric traits...... 97 Table 3-8. Significant SNPs (FDR < 0.1) and closest P. trichocarpa genes discovered through GWAS for arthropod community composition multitrait...... 98

vii Table 3-9. Candidate genes for mediating biotic interactions selected for functional analysis from networks. Breadth indicates number of connected network layers, depth refers to the number of neighbor genes in network connected to anchor gene, and edges indicate the number of connections (co-methylation, co-expression, metabolite GWAS and pyMBMS GWAS) among genes in networks...... 99 Table 3-10. Genome function enrichment analysis of co-expressed neighbor genes from network analysis. Significance of enrichment was determined based on Fisher Exact p-value with multiple testing correction to meet a threshold of FDR < 0.1...... 100 Table 4-1. NMDS and ANOSIM (Bray-Curtis dissimilarity) results for multivariate floral visitor composition tested against sex grouping (male vs female). p-values < 0.05 (bolded) indicate that floral visitor composition is more similar within replicate observations of sex group rather than among all observations...... 136 Table 4-2. NMDS and ANOSIM (Bray-Curtis dissimilarity) results for multivariate chemistry composition tested against sex grouping (male vs female). p-values < 0.05 (bolded) indicate that chemistry composition is more similar within replicate observations of sex group rather than among all observations...... 137 Table 4-3. Mantel test results comparing pairwise Bray-Curtis dissimilarity matrices among insect communities and chemistry composition of flowers and leaves. Bolded p-values and positive rm indicate a significant test, suggesting that similarity in chemistry composition relates to similarity in insect community assemblage...... 138 Table 4-4. Environmental fit results for 2019 floral community NMDS analysis. Bolded p-values (<0.05) indicate a significant correlation of the independent variable with the NMDS configuration...... 139 Table 4-5. Nested ANCOVA model results for 2019 analysis of individual native bee abundances. Bolded p-values (<0.05) indicate significant model effects. y ~ Sex + Location + Tree(Sex)&Random + Julian date ...... 140 Table 4-6. Nested ANCOVA model results for 2019 analysis of calculated community metrics. Bolded p-values (<0.05) indicate significant model effects. y ~ Sex + Location + Tree(Sex)&Random + Julian date ...... 141 Table 4-7. Test of random effects p-values extracted from nested ANCOVA for most abundant floral visitors as well as calculated species richness and Shannon Weaver diversity for 2019 survey analysis. Bolded values indicate that the independent variable had a significant effect on the dependent variable (p-value < 0.05)...... 142 Table 4-8. Environmental fit results for floral community NMDS analysis across years (2017- 2019). Bolded p-values (<0.05) indicate a significant correlation of the independent variable with the NMDS configuration...... 143 Table 4-9. Nested ANCOVA model results for year analysis. Bolded p-values (<0.05) indicate significant model effects. y ~ Sex + Year + Tree(Sex)&Random + Julian date ...... 144

viii Table 4-10. Test of random effects p-values extracted from nested ANCOVA for most abundant floral visitors as well as calculated species richness and Shannon Weaver diversity. Bolded values indicate that the independent variable had a significant effect on the dependent variable (p-value < 0.05)...... 145 Figure 2-1. Biotic phenotype symptoms observed on trees including (a) leaf symptoms of Melampsora sp. fungal leaf rust, (b) leaf spot symptoms of the Sphaerulina sp. fungus, (c) S. musiva canker symptoms, (d) branch gall created by the M. vagabunda aphid (e) petiole gall created by the P. populitransversus aphid, and (f) Phyllocolpa sp. leaf folding gall...... 58 Figure 2-2. Collinear genes in P. deltoides (Pd) and P. trichocarpa (Pt) chromosomes, based on MCScanX alignments. (a) Chr02; (b) Chr05...... 59 Figure 2-3. Change in number of tandemly duplicated genes discovered with increasing window size...... 60 Figure 2-4. QTL interval plots showing peaks across the genome that associate with 2008 biotic surveys. Lines on the plots indicate p-value thresholds as determined by running 1000 permutations of mapping model for (a) Melampsora sp. model with all individuals and clones subset to exclude individuals with S. musiva infection and (b) S. musiva leaf spot model with all individuals and clones subset to exclude individuals with Melampsora sp. infection...... 61 Figure 2-5. QTL interval plots showing peaks across the genome that associate with biotic surveys. Lines on the plots indicate p-value thresholds as determined by running 1000 permutations of mapping model for (a) binary fungal survey of S. musiva canker, (b) binary insect surveys of M. vagabunda and P. populitransversus, (c) and insect surveys of Phyllocolpa sp. leaf gall counts and overlapping peaks for gentisyl alcohol 5-O-glucoside compound QTL. .62 Figure 2-6. Mean and standard error of fungal infection for individuals with varying category levels of competing fungus infection. Letters indicate significantly different means as determined by Tukey's honest significance test for each one-way ANOVA test...... 63 Figure 2-7. Comparison of gene content in P. trichocarpa grandparent 93-968 (left line) and P. deltoides grandparent Ill-101 (right line) for significant genetic intervals for (a) Melampsora sp. chromosome 4, (b) S. musiva chromosome 16, (c) M. vagabunda chromosome 5, (d) Phyllocolpa sp. chromosome 10, and (e) Phyllocolpa sp. chromosome 13 associations. QTL intervals were defined as 1 Mb regions centered on the marker with the highest LOD score. Size of gene point is relative to the number of genes in the tandem duplication expansion...... 64 Figure 3-1. Map displaying genotype origin populations and survey sites – circles represent collection locations in the wild, stars represent common gardens, the survey sites...... 103 Figure 3-2. Arthropod community NMDS plot depicting 3 dimensions within all gardens surveyed in 2012. Color groupings indicate tree surveyed in specified garden...... 104 Figure 3-3. Arthropod community NMDS plots depicting 3 dimensions for (a) 2012 Clatskanie garden survey and (b) 2012 Placerville garden survey. PERMANOVA analysis, run with only genotypes with replicate observations, output indicated for each garden in the upper right corner

ix of each plot for each common garden. Size of species text indicates position relative to the dimensional space, for example, larger font indicates forward projection along the positive values of axes...... 105 Figure 3-4. Arthropod community NMDS plots depicting 3 dimensions for (a) 2012 Corvallis garden survey and (b) 2015 Corvallis garden survey. PERMANOVA analysis, run with only genotypes with replicate observations, output indicated for each garden in the upper right corner of each plot for each common garden. Size of species text indicates position relative to the dimensional space, for example, larger font indicates forward projection along the positive values of axes...... 106 Figure 3-5. Manhattan plot output from GWAS for (a) individual arthropods (single-trait; three tests), (b) calculated community metrics for each garden (single-trait; three tests), and (c) community composition (multi-trait; two tests). SNPs that passed the red dotted line were significantly associated with indicated trait. Clatskanie community (c) did not pass threshold but contained multiple suggestive markers...... 107 Figure 3-6. Co-expression network for Clatskanie Phyllonorycter sp. single-trait gene candidates. Co-expression genes were grouped based on their biological function as determined by Gene- ontology terms...... 108 Figure 3-7. Co-expression network for Corvallis arthropod richness single-trait gene candidates. Co-expression genes were grouped based on their biological function as determined by Gene- ontology terms...... 109 Figure 3-8. Co-expression network for Corvallis community composition multi-trait gene candidates. Co-expression genes were grouped based on their biological function as determined by Gene-ontology terms...... 110 Figure 4-1. Map of all thirty Salix nigra tree locations in WVU Core Arboretum. Pin shapes indicate tree selected for floral insect community survey. Points indicate additional individuals in the Arboretum...... 149 Figure 4-2. Tree canopy pan-trap design. Camouflaged buckets were added to hang below river traps to add weight, preventing winds from jostling traps...... 150 Figure 4-3. Non-metric multidimensional plot (dimensions = 3; stress = 0.12) of insect floral community with groupings indicated by color for sex of tree (ANOSIM R =0.3077, p-value = 0.001) for 2019 analysis. Size of species text indicates position relative to the dimensional space, for example, larger font indicates forward projection along the positive values of all axes...... 151 Figure 4-4. Non-metric multidimensional plot (dimensions = 4; stress = 0.12) of 2017-2019 insect floral community with groupings indicated by color for sex of tree (ANOSIM R = 0.1232, p-value = 0.011). Size of species text indicates position relative to the dimensional space, for example, larger font indicates forward projection along the positive values of all axes...... 152

Figure 4-5. Total concentration averages of catkin metabolites, leaf metabolites and catkin VOCs for male and female trees. Letters to the left of boxes indicate significantly different means (p- value < 0.05) determined by one-way ANOVA test...... 153 Figure 4-6. Total concentration averages of monoterpenes and sesquiterpenes VOCs for male and female trees. Letters to the left of boxes indicate significantly different means (p-value < 0.05) determined by one-way ANOVA test...... 154 Figure 4-7. (a) Non-metric multidimensional plot (dimensions = 3; stress = 0.12) of insect floral community with groupings indicated by color for survey method (ANOSIM R = 0.6656, p-value = 0.001). (b) Breakdown of insect capture/observation for two canopy survey methods. Additional pie charts around pan-trap pie chart indicate the percentage of that order captured in different colors of pan-traps...... 155 Figure 4-8. Average activity of most common floral visitors and calculated community metrics from 2019 S. nigra surveys. Letters to the left of boxes indicate significantly different means as determined by a Tukey’s HSD (p-value < 0.05)...... 156 Figure 4-9. Correlation plots of A. nigrae with VOC compounds...... 157 Figure 4-10. Average values of species richness and Shannon-Weaver diversity for female and male trees from year analysis model. Letters to the left of boxes indicate significantly different means as determined by a Tukey’s HSD (p-value < 0.05) for each separate nested ANCOVA model...... 158 Figure 4-11. (a) Amount of time and date in each year (2017, 2018, and 2019) in which catkins were actively attractive to insects. End date for male individuals indicates trees have dropped all catkins while female trees no longer have receptive stigmas (brown and shriveled) or active insect in canopies. (b) Average count of A. nigrae visits to S. nigra catkins in survey years 2017, 2018 and 2019 from year analysis model. Letters to the left of boxes indicate significantly different means as determined by a Tukey’s HSD (p-value < 0.05)...... 159

Chapter 1: Introduction: Biotic interactions in the Salicaceae

1 1.1 Community genetics

The average person often overlooks insects and fungi, but it is difficult to ignore the impact of these organisms on our planet. In agricultural systems it is estimated that 16-18% of all crop losses globally are due to insect and fungal pathogens and, in an attempt to mitigate attack, over $13 billion was spent on pesticide application in 2004 in the United States alone (Oerke, 2006).

Additionally, in forests in 2015, more than 6 million acres of trees were destroyed by herbivores

(Karel & Man, 2017), a problem that has the potential to grow as climate change progresses

(Garrett et al., 2006). Despite the cost associated with detrimental organisms, insects also provide over $57 billion in ecological services associated with human recreation such as education, aesthetic enjoyment, and gardening in the United States (Zhang et al., 2007; Mace et al., 2012).

Furthermore, plants receive a number of useful services from insects including pest control, nutrient cycling, pollination and seed dispersal (Losey & Vaughan, 2006; Zhang et al., 2007; Potts et al.,

2010). It is estimated that over 80% of wild plant productivity is reliant upon insects for seed and fruit set (Kremen et al., 2007; Potts et al., 2010). Research focused on the biotic-host plant relationship is imperative to understanding and managing ecosystems to balance the harmful interactions and beneficial services that insects provide to both humans and plants.

To identify a host, insects undergo a series of sequential behaviors including locating habitat, locating a host, assessing and accepting an individual host, and finally feeding. Host location is of particular importance in insect-plant interactions and can be driven by any combination of physiological and morphological cues (Panda & Khush, 1995). Although the host selection process seems simple, natural ecosystems are complex as they are composed of interacting assemblages of plant, fungal, insect, and animal species. One method for disentangling the organization of these relationships begins at the plant genetic level. In natural plant populations there is often genetic

2 variation in characteristics that attract or deter biotic association (Crutsinger et al., 2006). If the variation in a trait is strongly heritable, plants are able to extend their phenotypes and act as a mediator of multi-biotic interactions which can in turn structure whole communities of organisms

(Whitham et al., 2006). Determining how shifts in genes contribute to the biotic-plant relationship effectively links ecology and evolution, and allows for prediction of community level responses to changing conditions (Shuster et al., 2006; Allan et al., 2013).

1.2 The Salicaceae family as a genetic model

The Salicaceae family, which includes aspens/cottonwoods (Populus) and willows (Salix), contains trees and shrubs that are of both agricultural and ecological importance. With rapid growth and capability for clonal reproduction, many species of Populus and Salix have become a focus for research into biofuel production making them important commercial crops (Rubin, 2008; Hinchee et al., 2011; Langholtz et al., 2016). Additionally, Populus and Salix are considered foundational and dominant species in riparian environments elevating their impact on natural ecosystems

(Whitham et al., 2006; Crutsinger, 2016). Finally, the completion of full genome sequencing of both Populus trichocarpa and Populus deltoides has made the family an important genetic model for ecological research of forest trees (Tuskan et al., 2006). Over the last decade, Populus has been utilized to explore the genetic relationship between plants and communities of insects and animals.

Many species of Populus have overlapping distributions where they form naturally occurring hybrid zones that have been a target for community assemblage research (Stettler et al., 1996;

Floate et al., 2016). Studies of these zones have shown that tree genetic composition can both directly and indirectly influence community interactions. For example, hybrid zones have been found to directly mediate cryptic speciation of arthropods such as the mite Aceria parapopuli

(Evans et al., 2008). Similarly, beaver browsing of particular Populus genotypes modifies

3 palatability of plant tissue which indirectly influences Chrysomela confluens beetle feeding

(Martinsen et al., 1998; Bailey et al., 2004).

Despite the number of papers demonstrating the community assembling potential of plant genotypes in hybrid Populus systems, only a few publications have emerged that identify the underlying genes important in these interactions in pure species and even fewer determine if the same community effects are detectable for insects that interact with floral tissue rather than leaf, wood, or soil (Crutsinger et al., 2014; Crutsinger, 2016; Barker et al., 2019). Additionally, a frequent observation of the multi-species Populus hybrid zones is that hybrids are more susceptible to insect/fungal attack and show more significant responses to community interactions than their pure species counterparts (Whitham et al., 1999). A possible explanation is that elevated community responses are the result of hybrid production of novel traits (or expression of traits) that are qualitatively or quantitatively different from the pure parental species (Rieseberg et al., 1993;

Cheng et al., 2011). Hybrids therefore can be a valuable tool in initial discovery of underlying loci and genetic structure influencing biotic distributions prior to determining if mechanisms exist in a pure species (Whitham et al., 1999).

1.3 Physiological and morphological features

There are several genetically controlled traits in Salicaceae that are candidates for mediating insect- plant and community level interactions. Secondary metabolites are one such physiological trait with large literature support for their importance in insect-plant relationships. Phenolics are the most plentiful secondary metabolites in the Salicaceae family (Freeman & Beattie, 2008), and their main function is to act as a feeding deterrent to generalist insects and animals (Lindroth & Peterson,

1988; Bailey et al., 2006; Boeckler et al., 2011). However, in certain cases specialist insects prefer

4 trees high in these defenses as they have either adapted to or utilize them for their own protection

(Pasteels et al., 1983; Kearsley & Whitham, 1989; Matsuki & MacLean, 1994). Secondary metabolites have also been determined to vary dramatically after hybridization. For example, across

30 studies and 1,112 secondary metabolites it was found that only 70.3% were present in both hybrids and their parents, whereas 24.3% were not present in hybrids and 5.5% of compounds were completely novel to the system (Cheng et al., 2011). Phenolics are also highly variable among pure species with some, such as salicin, found in many but others restricted to only a few (Boeckler et al., 2011). Similarly, simple genetically heritable morphological traits, such as leaf or petiole shape, can impact insect distributions and are highly variable among pure species and hybrids in the

Salicaceae (Whitham, 1989; Robinson et al., 2012).

The Salicaceae family also presents a unique opportunity to study the community impact of a relatively rare plant characteristic, that is only present in 6% of angiosperms: dioecy (Renner &

Ricklefs, 1995; Barrett & Hough, 2013). All species of Populus and Salix are dioecious, having separate male and female individuals. There have been several studies that considered the role that dioecy and sex differences in resource allocation can play on community level interactions. For example, male individuals in both S. lasiolepis and S. cinerea supported higher densities of tenthredinid sawflies than their female counterparts (Alliende, 1989; Boecklen et al., 1990).

Additionally, Phratora vulgatissima prefers to feed on female hosts of S. cinerea despite higher egg mortality associated with higher predator presence on female plant (Kabir, 2012). However, these studies have focused solely on leaf feeding herbivores, thereby neglecting the floral visitor community, which is of great potential importance to the fitness of plant species in the family.

1.4 Dioecy and floral visitor communities

5 Although not a trait common to Populus, there are many species of Salix that rely heavily upon communities of insects to carry pollen among flowers for successful sexual reproduction (Tollsten

& Knudsen, 1992; Füssel et al., 2007; Ashman, 2009). Salix species rely on a complex combination of volatile organic compounds as olfactory signals, and male flowers are often yellow in color which acts as additional visual stimuli that female flowers lack (Tollsten & Knudsen, 1992; Füssel et al., 2007). Salix floral resources are also extremely important in supplying food for many native herbivores, as flowers both act as a nutritive resource sink and offer rewards such as nectar and pollen (Ostaff et al., 2015).

Work that has been done on Salix floral communities has almost exclusively focused on bee interactions while neglecting the diverse groups of insects that are also known to visit flowers such as flies, beetles, and aphids (Dötterl et al., 2005; Füssel et al., 2007; Ostaff et al., 2015). Many of these papers also focus closely on volatiles and have not considered additional characteristics such as floral chemical defenses in insect distributions. The potential interactions and competition between bees and other floral herbivores could have major implications for both the plant sexual reproduction and bee survival. With the current global declines in bee species populations (Potts et al., 2010) understanding how species like Salix attract and preserve resources for their cross- pollinators, despite the potential diversity of insects interacting with floral resources, is an important area of research that remains relatively unexplored.

6 1.5 Objectives

Building on the community structuring potential of the host plant genome that has already been established in Populus and Salix, I studied the host genetic influence on biotic communities by addressing three primary objectives in this dissertation:

Objective 1 (Chapter 2): Identify and understand the genetic mechanisms of fungal and insect association in Populus. This objective was accomplished by using a hybrid family cross of P. trichocarpa x P. deltoides and quantitative trait loci analysis (QTL) as a tool to identify genetic regions of interest. I then compared the QTL interval composition among the parental genomes to look for structural and functional enrichments to identify candidate genes important to biotic-biotic interactions. This manuscript has now been published in the journal Ecology and Evolution (Simon et al. 2020).

Objective 2 (Chapter 3): Identifying genes underlying arthropod community composition in a pure species population of Populus trichocarpa. I accomplished this objective by surveying arthropods in three common gardens containing over 1,000 genotypes from across the species range and using both single-trait and multi-trait genome-wide-association analysis (GWAS) and functional networks to identify SNPs associated with abundant insects and community composition.

Objective 3 (Chapter 4): Determining the impact of dioecy in Salix on assembly of floral insect communities and activity of native bee species. This objective was accomplished by visually surveying insects that visited flowers of the tree form willow Salix nigra across three years.

Additionally, I utilized pan-trapping techniques to compare to visual surveys and capture a fuller community estimate in the last year of surveys. This coupled with non-metric multidimensional

7 scaling (NMDS) analysis and Mantel tests allowed me to assess the effects of tree sex, floral scent, defensive chemistry and inter-annual variability of flower emergence on distributions of floral visitors and primary cross-pollinators important in sexual productivity. This manuscript has been submitted to Oecologia.

8 1.6 Literature cited

Allan GJ, Shuster SM, Woolbright S, Walker FM, Meneses N, Keith A, Bailey JK, Whitham

TG. 2013. Interspecific indirect genetic effects (iiges): Linking genetics and genomics to

community ecology and ecosystem processes. In: Ohgushi T, Schmitz O, Holt RD, eds.

Trait-Mediated Indirect Interactions Ecological and Evolutionary Perspectives. Cambridge,

UK: Cambridge University Press, 295–323.

Alliende MC. 1989. Demographic studies of a dioecious tree. II. The distribution of leaf predation

within and between trees. The Journal of Ecology 77: 1048–1058.

Ashman TL. 2009. Sniffing out patterns of sexual dimorphism in floral scent. Functional Ecology

23: 852–862.

Bailey JK, Schweitzer JA, Rehill BJ, Lindroth RL, Martinsen GD, Whitham TG. 2004.

Beavers as molecular geneticists: a genetic basis to the foraging of an ecosystem engineer.

Ecology 85: 603–608.

Bailey JK, Wooley SC, Lindroth RL, Whitham TG. 2006. Importance of species interactions to

community heritability: a genetic basis to trophic‐level interactions. Ecology Letters 9: 78–

85.

Barker HL, Riehl JF, Bernhardsson C, Rubert‐Nason KF, Holeski LM, Ingvarsson PK,

Lindroth RL. 2019. Linking plant genes to insect communities: Identifying the genetic

bases of plant traits and community composition. Molecular Ecology 28: 4404–4421.

Barrett SC, Hough J. 2013. Sexual dimorphism in flowering plants. Journal of Experimental

Botany 64: 67–82.

Boecklen WJ, Price PW, Mopper S. 1990. Sex and drug and herbivores: Sex-biased herbivory in

arroyo willow (Salix lasiolepis). Ecology 71: 581–588.

Boeckler GA, Gershenzon J, Unsicker SB. 2011. Phenolic glycosides of the Salicaceae and their

9 role as anti-herbivore defenses. Phytochemistry 72: 1497–1509.

Cheng D, Vrieling K, Klinkhamer PGL. 2011. The effect of hybridization on secondary

metabolites and herbivore resistance: implications for the evolution of chemical diversity in

plants. Phytochemistry Reviews 10: 107–117.

Crutsinger GM. 2016. A community genetics perspective: opportunities for the coming decade.

New Phytologist 210: 65–70.

Crutsinger GM, Collins MD, Fordyce JA, Gompert Z, Nice CC, Sanders NJ. 2006. Plant

genotypic diversity predicts community structure and governs an ecosystem process.

Science 313: 966–968.

Crutsinger GM, Rodriguez-Cabal MA, Roddy AB, Peay KG, Bastow JL, Kidder AG, Dawson

TE, Fine PV, Rudgers JA. 2014. Genetic variation within a dominant shrub structures

green and brown community assemblages. Ecology 95: 387–398.

Dötterl S, Füssel U, Jürgens A, Aas G. 2005. 1,4-Dimethoxybenzene, a floral scent compound in

willows that attracts an oligolectic bee. Journal of Chemical Ecology 31: 2993–2998.

Evans LM, Allan GJ, Shuster SM, Woolbright SA, Whitham TG. 2008. Tree hybridization and

genotypic variation drive cryptic speciation of a specialist mite herbivore. Evolution:

International Journal of Organic Evolution 62: 3027–3040.

Floate KD, Godbout J, Lau MK, Isabel N, Whitham TG. 2016. Plant–herbivore interactions in a

trispecific hybrid swarm of Populus: assessing support for hypotheses of hybrid bridges,

evolutionary novelty and genetic similarity. New Phytologist 209: 832–844.

Freeman BC, Beattie GA. 2008. An overview of plant defenses against pathogens and herbivores.

The Plant Health Instructor.

Füssel U, Dötterl S, Jürgens A, Aas G. 2007. Inter- and intraspecific variation in floral scent in

the genus Salix and its implication for pollination. Journal of Chemical Ecology 33: 749–

10 765.

Garrett KA, Dendy SP, Frank EE, Rouse MN, Travers SE. 2006. Climate change effects on

plant disease: genomes to ecosystems. Annual Review of Phytopathology 44: 489–509.

Hinchee M, Rottmann W, Mullinax L, Zhang C, Chang S, Cunningham M, Pearson L, Nehra

N. 2011. Short-rotation woody crops for bioenergy and biofuels applications. In: Biofuels.

New York, NY: Springer New York, 139–156.

Kabir M. 2012. The influence of plant sex on the performance of a detrimental herbivore and two

biocontrol agents in the dioecious grey willow, Salix cinerea.

Karel T, Man G. 2017. Major forest insect and disease conditions in the United States: 2015.

Washington, D.C.

Kearsley MJ, Whitham TG. 1989. Developmental changes in resistance to herbivory:

implications for individuals and populations. Ecology 70: 422–434.

Kremen C, Williams NM, Aizen MA, Gemmill-Herren B, LeBuhn G, Minckley R, Packer L,

Potts SG, Roulston T, Steffan-Dewenter I, et al. 2007. Pollination and other ecosystem

services produced by mobile organisms: a conceptual framework for the effects of land-use

change. Ecology Letters 10: 299–314.

Langholtz MH, Stokes BJ, Eaton LM. 2016. 2016 Billion-ton report: Advancing domestic

resources for a thriving bioeconomy, Volume 1: Economic availability of feedstock (Oak

Ridge National Laboratory, Ed.). Oak Ridge, TN (United States): UT-Battelle, LLC for the

US Department of Energy.

Lindroth RL, Peterson SS. 1988. Effects of plant phenols of performance of southern armyworm

larvae. Oecologia 75: 185–189.

Losey JE, Vaughan M. 2006. The economic value of ecological services provided by insects.

Bioscience 56: 311–323.

11 Mace GM, Norris K, Fitter AH. 2012. Biodiversity and ecosystem services: a multilayered

relationship. Trends in Ecology & Evolution 27: 19–26.

Martinsen GD, Driebe EM, Whitham TG. 1998. Indirect interactions mediated by changing

plant chemistry: beaver browsing benefits beetles. Ecology 79: 192–200.

Matsuki M, MacLean SF. 1994. Effects of different leaf traits on growth rates of insect herbivores

on willows. Oecologia 100: 141–152.

Oerke EC. 2006. Crop losses to pests. The Journal of Agricultural Science 144: 31–43.

Ostaff DP, Mosseler A, Johns RC, Javorek S, Klymko J, Ascher JS. 2015. Willows (Salix spp.)

as pollen and nectar sources for sustaining fruit and berry pollinating insects. Canadian

Journal of Plant Science 95: 505–516.

Panda N, Khush GA. 1995. Host plant resistance to insects. Wallingford, UK: CAB International.

Pasteels JM, Rowell‐Rahier M, Braekman JC, Dupont A. 1983. Salicin from host plant as

precursor of salicylaldehyde in defensive secretion of chrysomeline larvae. Physiological

Entomology 8: 307–314.

Potts SG, Biesmeijer JC, Kremen C, Neumann P, Schweiger O, Kunin WE. 2010. Global

pollinator declines: trends, impacts and drivers. Trends in Ecology & Evolution 25: 345–

353.

Renner SS, Ricklefs RE. 1995. Dioecy and it correlates in the flowering plants. American Journal

of Botany 82: 596–606.

Rieseberg LH, Ellstrand NC, Arnold M. 1993. What can molecular and morphological markers

tell us about plant hybridization? Critical Reviews in Plant Sciences 12: 213–241.

Robinson KM, Ingvarsson PK, Jansson S, Albrectsen BR. 2012. Genetic variation in functional

traits influences arthropod community composition in aspen (Populus tremula L.). PLoS

ONE 7: e37679.

12 Rubin EM. 2008. Genomics of cellulosic biofuels. Nature 454: 841–845.

Shuster SM, Lonsdorf E V., Wimp GM, Bailey JK, Whitham TG. 2006. Community

heritability measures the evolutionary consequences of indirect genetic effects on

community structure. Evolution 60: 991–1003.

Stettler R, Bradshaw T, Heilman P, Hinckley T. 1996. Biology of Populus and its implications

for management and conservation. Ottawa, ON, Canada: NRC Research Press.

Tollsten L, Knudsen JT. 1992. Floral scent in dioecious Salix (Salicaceae)- a cue determining

pollination system? Plant Systematics and Evolution 182: 229–237.

Tuskan GA, DiFazio S, Jansson S, Bohlmann J, Grigoriev I, Hellsten U, Putnam N, Ralph S,

Rombauts S, Salamov A, et al. 2006. The genome of black cottonwood, Populus

trichocarpa (Torr. & Gray). Science 313: 1596–1604.

Whitham TG. 1989. Plant hybrid zones as sinks for pests. Science 244: 1490–1493.

Whitham TG, Bailey JK, Schweitzer JA, Shuster SM, Bangert RK, LeRoy CJ, Lonsdorf E V.,

Allan GJ, DiFazio SP, Potts BM, et al. 2006. A framework for community and ecosystem

genetics: from genes to ecosystems. Nature Reviews Genetics 7: 510–523.

Whitham TG, Martinsen GD, Keim P, Floate KD, Dungey HS, Potts BM. 1999. Plant hybrid

zones affect biodiversity: tools for a genetic‐based understanding of community structure.

Ecology 80: 416–428.

Zhang W, Ricketts TH, Kremen C, Carney K. 2007. Ecosystem services and dis-services to

agriculture. Ecological Economics 64: 253–260.

Chapter 2: Host plant genetic control of associated fungal and insect

species in a Populus hybrid

In press: Simon S, Tschaplinski T, Leboldus J, Keefover-Ring K, Azeem M, Chen J, Macaya-Sanz D., MacDonald W., Muchero W., & DiFazio S. 2020. Host plant genetic control of associated fungal and insect species in a Populus hybrid. Ecology and Evolution. Authors’ contributions: S.S., S.D., and D.M.-S. identified, collected, and analyzed insect surveys. W.M., J.L., J.C., and W.M. identified and collected fungal species datasets. K.K.-R., M.A., and T.T. collected/analyzed metabolite data. S.S. and S.D. analyzed the data. W.M., G.T. and S.D. designed and supervised the study. S.S., S.D., J.L., T.T., and K.K.-R. wrote the manuscript.

2.1 Abstract

Plants employ a diverse set of defense mechanisms to mediate interactions with insects and fungi.

These relationships can leave lasting impacts on host plant genome structure, such as rapid expansion of gene families through tandem duplication. These genomic signatures provide important clues about the complexities of plant/biotic stress interactions and evolution. We used a pseudo-backcross hybrid family to identify Quantitative Trait Loci (QTL) controlling associations between Populus trees and several common Populus diseases and insects. Using whole genome sequences from each parent, we identified candidate genes that may mediate these interactions.

Candidates were partially validated using mass spectrometry to identify corresponding QTL for defensive compounds. We detected significant QTL for two interacting fungal pathogens and three insects. The QTL intervals contained candidate genes potentially involved in physical and chemical mechanisms of host-plant resistance and susceptibility. In particular, we identified overlapping

QTLs for a phenolic glycoside and Phyllocolpa sawfly abundance. There was also significant enrichment of recent tandem duplications in the genomic intervals of the parent that was native to the pest range, but not the exotic parent. Tandem gene duplication may be an important mechanism for rapid response to biotic stressors, enabling trees with long juvenile periods to reach maturity despite many coevolving biotic stressors.

2.2 Introduction

In natural ecosystems, the dynamics of plant interactions with other living organisms are complex.

This is especially true for organisms that rely on plants for shelter, nutrition, and reproduction, such as fungi and insects (Panda & Khush, 1995; Chisholm et al., 2006). Although fungi and insects can provide some of the same useful services in return, those that do not can be extremely harmful to plant productivity. To mitigate the effects of biotic stress, plants employ a diverse set of defense mechanisms, including chemical, protein-derived molecules, and physical barriers (Panda &

Khush, 1995). Insects and fungi must develop strategies in parallel to overcome these obstacles to survive (Mello & Silva-Filho, 2002; Chisholm et al., 2006). The theory of gene-for-gene coevolution has frequently been used to describe this host plant genetic relationship to its arthropod and fungal communities (Ehrlich & Raven, 1964; Thompson, 1988; Mello & Silva-Filho, 2002;

Chisholm et al., 2006).

The gene-for-gene theory suggests a very simple dynamic for the genetic interactions that occur between two species. A gene in the host plant that is important in biotic relationships has a corresponding, coevolving gene from a pathogen/insect, which can lead to resistance or susceptibility depending on the life history of the pathogen/insect (Flor, 1971; Friesen et al., 2007).

Much of the evidence for these interactions has been found in crop systems where plant species often have dominant, single-gene trait conferring resistance to feeding (Thompson, 1988). For example, there are over twenty different genes in wheat (Triticum aestivum L.) that each confer resistance to the Hessian fly, Mayetiola destructor (Thompson & Burdon, 1992). Exposure of

Hessian fly populations to these resistant varieties of wheat created selection pressure that led to increased virulent gene combinations in the pest (Gallun, 1977; Panda & Khush, 1995). Similarly, in plant-fungal systems, breeding for dominant resistance in cereal crops resulted in new selective

forces that increased virulence gene frequencies in Puccinia spp. cereal rusts (Pretorius et al., 2000;

Chen, 2005). This in turn can lead to an evolutionary arms race between plants, insects, and fungi with the continual development of mechanisms to overcome both genetic defenses and virulent attacks (Thompson & Burdon, 1992; Bergelson et al., 2001).

The relationships of host plant genetics and biotic associations can also be more complex than these crop breeding systems suggest, and they can leave a lasting impact on genome structure (Lefebvre

& Chèvre, 1995). Host plant and biotic associations can lead to the expansion of gene families responsible for the host plant response to biotic stress. For example, the Kunitz protease inhibitors

(KPIs) in Populus are important in defense responses against insects by inhibition of herbivore digestion (Haruta et al., 2001; Major & Constabel, 2008). The KPI gene family has greatly expanded in response to insect attack through tandem duplication events (Philippe et al., 2009).

Similarly, plant resistance (R) genes, which are important in the defense response of plants to fungal pathogen attack, have also expanded through tandem and segmental duplication events due to biotic pressures (Hulbert et al., 2001; Leister, 2004). Analyzing how the genome is structured in the host plant when it associates with fungi and insects is important for studying these relationships and understanding the complexities of their genetic interactions.

Given their rapid growth and vegetative reproduction, Populus species have become a focus for research into biofuel production, making them a valuable commercial crop (Meilan et al., 2002;

Taylor, 2002; Stanton et al., 2010). Populus has also become an important genetic model for research into a wide variety of ecological and adaptive traits (McKown et al., 2014), including interactions with the biotic community (Whitham et al., 2006; Crutsinger et al., 2014). In particular, interspecific hybrids of P. trichocarpa x P. deltoides segregate for a wide variety of

traits, including resistance to insect and fungal attack (Newcombe, 1998; Newcombe & Ostry,

2001). Such hybrid family crosses can be used to identify regions of the genome that are important in mediating biotic stress.

In this study, we investigated the genome composition of loci associated with insect and fungal species in an interspecific Populus family. We surveyed insects and fungal pathogens in a P. deltoides x P. trichocarpa pseudo-backcross family and used quantitative trait locus (QTL) analysis and comparative genomics to address three main questions: (1) Is there heritable, host genetic control of fungal and insect species? (2) What protein domains and gene ontology terms are enriched in the QTL intervals in the genomes of each Populus species? (3) What candidate genes in the intervals are unique to each species when comparing the P. trichocarpa and P. deltoides genomes?

2.3 Methods

QTL mapping pedigree

The 52-124 family was developed by crossing a male Populus deltoides, ILL-101 from southern

Illinois, with a female Populus trichocarpa clone, 93-968 from western Washington State. The F1 clone, 52-225, was then crossed with a male Populus deltoides clone, D124 from Minnesota, to generate the final 749 progeny in the pseudo-backcross population.

Field site descriptions

Clonal cuttings of the 52-124 progeny were obtained from the University of Minnesota and the

University of Florida. The individuals were propagated in the West Virginia University (WVU) greenhouses in March 2006. A hay production field was tilled and disked at the WVU Agronomy

farm (39°39′32″N 79°54′19″W) in Morgantown, West Virginia for planting. In July 2007 the rooted cuttings were planted with two clonal replicates for each of 749 genotypes at 2 m x 2 m spacing in an interlocking block design. The plantation was thinned in December of 2008 by removing 50% of the trees in a diamond fashion resulting in 2.83 m x 4 m spacing of the remaining trees. A plantation of the same family cross was established at the Westport Research Station in the

Columbia River floodplain (46°07'58.8"N 123°22'05.0"W) in April, 2016. The trial was planted with a total of 339 progeny replicated in a randomized complete block design with three blocks.

SNP genotyping and genetic map construction

SNP loci were selected from whole genome resequencing data generated for the parent trees of the pedigree, focusing on loci that were fixed for different alleles in P. deltoides and P. trichocarpa.

Sequencing was performed on the Illumina GAII system with single end read lengths of 75 bp and a total depth of ~35X on average. Reads were aligned using MAQ and SNPs were called using

Samtools mpileup with a minimum quality of 30 and a minimum depth of 5 reads per allele, and a subset of loci were confirmed by Sanger sequencing (Slavov et al., 2012). Polymorphic loci were selected that were maternally informative for P. trichocarpa. These were incorporated into an

Illumina Infinium Bead Array, which was used to genotype 3,568 SNP loci in 692 of the progeny.

This data was then used to create the genetic map composed of 19 linkage groups corresponding to

19 Populus chromosomes. Genotyping and map construction are described in more detail elsewhere

(Muchero et al., 2015).

Family 52-124 parent and progeny phenotyping

In order to identify regions of the Populus genome associated with biotic stress, a variety of fungal pathogens and insect herbivore species were surveyed in the WVU Morgantown and Westport

Oregon plantation sites. Melampsora sp. rust was identified visually by local pathologist Dr.

William MacDonald. Sphaerulina musiva was identified by sequencing of the ITS region (Verkley et al., 2013). Insect identification was completed using insect morphological features, feeding symptoms, and known hosts/species distributions for Phyllocolpa spp. (Smith & Fritz, 1996;

Kopelke, 2007), Mordwilkoja vagabunda (Ignoffo & Granovsky, 1961a,b), and Pemphigus populitransversus (Bird et al., 1979; Faith, 1979).

Melampsora sp. leaf rust (Figure 1a) and Sphaerulina musiva fungal leaf spot symptoms (Figure

1b) were scored on a 0-3 scale of disease severity, with 0 indicating absence of symptoms and 3 indicating high degree of pathogen leaf damage, in the 2008 growing season for all 1353 tree canopies. In the fall of 2014, stem canker symptoms caused by the same fungus Sphaerulina musiva (Figure 1c) were scored on a 0-5 disease severity scale for a subset 498 unique genotypes and a total of 580 trees. Upon further examination of field conditions for S. musiva disease severity, it was determined that none of the progeny displayed complete resistance. The original 0-5 scale was binarized with scores from 2.5-5 scaled to 1 and 0-2 scaled to 0. The new scale indicated the progression of infection, with 1 specifying severe canker development and 0 indicating less aggressive canker symptoms.

Two abundant species of galling aphid were also surveyed during the 2016 growing season. To equalize the biomass surveyed for the aphid insect observations on each tree, branches of equal diameter were selected for insect counts and the remaining branches in the canopy were not surveyed. Tree canopy presence or absence for the petiole galling aphid

Pemphigus populitransversus (Figure 1d) and the branch galling aphid Mordwilkoja vagabunda

(Figure 1e) was recorded for 201 unique genotypes and a total of 218 trees. For all of the damage

scores in every survey year, 0 indicated a complete absence of fungal or insect presence in tree canopy.

In early August 2017 at the Westport site, full canopies were scored by counting galls of the leaf folding sawfly Phyllocolpa sp. (Figure 1f), on 534 unique genotypes and a total of 1020 trees. To estimate productivity of individuals and confirm that availability of resources did not drive insect attraction or feeding, main stem diameter in millimeters was recorded for all trees to be used as a covariate in the analysis.

Leaf metabolites

The metabolomic profiles for 211 genotypes of Family 52-124 were determined for leaves (leaf plastochron index 7 ± 2) sampled on September 14, 2006 at a USDA-FS field site near Grand

Rapids, MI from plants in their first growing season in the field. Leaves were quickly collected and flash-frozen on dry ice, prior to shipping back to Oak Ridge National Laboratory, where ~200 mg fresh weight per sample were extracted in 80% ethanol (aqueous), as described in Zhao et al.,

(2010) and analyzed for the major aromatic metabolites following trimethylsilylation and analysis by gas chromatography-mass spectrometry (GC-MS) using electron impact ionization (70 eV). The analytical and data extraction protocols were similar to those reported previously in Weston et al.,

(2015). The major phenolic metabolites quantified included salicortin, α-salicyloylsalicin, salicyl- salicylic acid-2-O-glucoside, salirepin (gentisyl alcohol 2-O-glucoside), gentisic acid 5-O- glucoside, 1,2,4-benzenetriol, and gentisyl alcohol 5-O-glucoside.

2.4 Statistical analysis

Within-family broad-sense heritability (H2) calculation

To prevent arbitrary score bias, Melampsora sp. and S. musiva values were normalized to have a mean of 0 and standard deviation 1 (score.transform function of the CTT R package) by applying the inverse of the cumulative distribution function of the normal distribution to the sample percentile score (Gianola & Norton, 1981). Surveys of the Melampsora sp. leaf rust, S. musiva leaf spot, and counts of the Phyllocolpa sp. galls had within-garden microsite variation removed using thin-plate spline regression (fields R package; Table 1). Residuals of the spatial correction were added to the mean of each survey dataset for each tree observation to spatially correct and rescale the values. These corrected values were then used to calculate the proportion of variance in the fungal and insect distributions that was due to genotype using a linear-mixed model (lmer function of the lme4 R package) with insect counts and fungal scores as the response, tree genotype as the predictor, and either competing fungus score or stem diameter biomass estimates as a covariate where applicable. Broad-sense heritability was calculated as = /( + ), where is the 2 2 2 2 2 𝑔𝑔 𝑔𝑔 𝑒𝑒 𝑔𝑔 genetic variance due to genotype and is the residual variance.𝐻𝐻 Rapid,𝛿𝛿 simulation𝛿𝛿 𝛿𝛿 -based 𝛿𝛿exact 2 𝑒𝑒 likelihood ratio tests were used to evaluate𝛿𝛿 the significance of variation due to genotype for each linear model (exactRLRT function of the RLRsim R package). SAS software version 9.4 (2013) was used to test for the normality of all data sets and for all subsequent statistical analyses, and transformations were conducted when necessary. Finally, best linear unbiased predictors (BLUPs) were extracted from these models to use in the QTL analysis.

Quantitative trait loci (QTL) analysis

The R software package R/qtl (Broman et al., 2003) was used for all QTL analyses described below. Composite interval mapping (CIM) was used to associate Melampsora sp. leaf rust, S. musiva leaf spot, and Phyllocolpa sp. to QTL positions (cim function). An additional CIM QTL analysis was conducted for fungi surveyed in 2008 to further evaluate potential competitive bias of

co-occurring fungal pathogens. This was done by sub-setting individuals out of the full dataset to exclude trees with symptoms of both pathogens, leaving us with 90 individuals that only showed

Melampsora sp. symptoms and 434 individuals only infected with S. musiva leaf spot. Single QTL mapping was used to associate the binary scores for the S. musiva canker, M. vagabunda, and P. populitransversus to QTL positions (scanone function). The method used for both mapping approaches was the expectation-maximization (EM) algorithm. Estimation of QTL interval significance was completed by performing 1000 permutations. Intervals with logarithm of odds

(LOD) scores that were above the p-value threshold (alpha = 0.05), as determined from the permutation tests, were selected for further analysis. The percent variance explained by significant markers for fungal and insect surveys, that were mapped using CIM, were calculated by extracting significant marker positions and creating a fit QTL model (fitQTL function). The positive allele contributing to an increase in susceptibility to fungi and insects was found by generating effect plots for each phenotype and its significant marker position (effectplot function).

Physical genome intervals

Physical genome intervals in the P. trichocarpa genome (v3.0) were examined for each significant

QTL for biotic associations. The intervals were defined as 1 Mb regions centered on the marker with the highest LOD score. Fixed physical genome sizes were used rather than intervals defined based on LOD scores due to the large variation in magnitude of LOD observed for the significant

QTL. For example, intervals of 1 LOD centered on the QTL ranged in size from 169 to 4620 kb.

Much of this variation was likely due to variation in marker density and local recombination rates, in addition to phenotyping and genotyping error. We believe that a fixed 1 Mb interval is a more consistent and conservative approach given the size of the family and the variation in strength of

the QTL (Yin et al., 2004b). On average, this represents approximately 6.34 cM, based on a total map size of 2617 cM and a total assembled genome length of 420 Mb.

Orthologous intervals were identified in the P. deltoides clone WV94 reference genome (v2.1) obtained from Phytozome (Goodstein et al., 2012). Orthology was determined using a combination of protein sequence conservation and synteny using MCScanX (Wang et al., 2012). Briefly, all proteins were compared in all-vs-all searches using blastp both within genomes and between genomes. These were then chained into collinear segments using the MCScanX algorithm.

Orthologous segments were identified based on the presence of large numbers of gene pairs in collinear order with high sequence identity (median blastp E score <1e-180) (Figure 2).

Synonymous (Ks) and nonsynonymous (Ka) nucleotide substitution rates were calculated using the

Bioperl DNAstatistics module (Stajich, 2002). Domain composition and Gene Ontology (GO) terms were obtained for each genome from Phytozome (v12.1). Intervals were customized for the grandparents of the pseudo backcross progeny (clones 93-968 for P. trichocarpa and D124 for P. deltoides) by converting the respective reference genome based on alignment of short-read sequences derived from each species. Specifically, we generated 243 and 248 million 250 bp paired end Illumina HiSeq sequences for 93-968 and D124, respectively. This yielded an average coverage of ~150X per genome. These were aligned to the respective reference genome for each species

(Nisqually v3.0 for P. trichocarpa, and WV94 v2.0 for P. deltoides) using bwa mem with default parameters. SNPs and small indels were identified using samtools mpileup and bcftools call with default multiallelic variant settings (Li et al., 2009; Li, 2011), and sequence depth was extracted using vcftools (Danecek et al., 2011). Sequences were converted using the vcftools utility vcf- consensus. Genes with no coverage in the alignments were excluded from the intervals for each species.

Tandem duplications

Tandemly duplicated genes were identified using all-vs-all blastp searches within each genome for biotic stress-associated intervals. Genes with blastp E scores < 1e-180 that were located within 500 kb of one another were considered to be recent tandem duplications. These were considered recent because P. trichocarpa and P. deltoides orthologs also have a median blastp E score in this range, suggesting that these tandem duplications mostly occurred after these species diverged from a common ancestor. The window size was determined by testing a range of values and choosing a window size at which the number of newly discovered tandem duplicates began to decline (Figure

3). The stringent E score cutoff was intended to focus the analysis on genes that are recently duplicated and, therefore, potentially differentially duplicated between the species. The QTL intervals were tested for significant enrichment of tandem duplicates by using a Monte Carlo simulation. Sets of contiguous genes equal in number to those contained in each QTL interval were randomly selected from the whole genome, and the number of sampled tandem duplications was counted for each iteration. This was repeated 10,000 times, and the observed number of tandem duplicates was compared to the simulated distribution to derive an empirical P-value.

2.5 Results

Heritability of fungal and insect associations

Clonal repeatability (or within-family broad-sense heritability) was estimated for each categorical survey trait (Table 1). There was a significant host-plant genetic contribution of moderate effect controlling the association of S. musiva leaf spot disease severity (H2 = 0.250, p-value < 0.0001).

There was a strong effect of host-plant genetics on associations of Melampsora sp. disease severity

(H2 = 0.609, p-value < 0.0001) and the continuous count of leaf-folding galls of Phyllocolpa sp. (H2

= 0.391, p-value < 0.0001).

QTL mapping of fungal and insect surveys

Four QTL intervals containing a total of 38 markers were significantly associated with the abundance of fungal pathogens (Table 2). An overlapping QTL on Chr04 (Figure 4a,b) was associated with both Melampsora sp. (marker position = 3.50687, p-value = 0.001) and S. musiva leaf spot disease severity (marker position = 15.4912824, p-value = 0.026) for analysis that included all clones. The marker with the highest LOD score explained 54.1% of the variance in the disease severity of Melampsora sp. within the Chr04 QTL interval, whereas the top marker within the overlapping interval only explained 3.31% of the variance in the disease severity of S. musiva in the analysis with all individuals.

Upon exclusion of individuals that showed symptoms of the competing fungus, the strength of the association on Chr04 increased for Melampsora sp. (marker position marker position = 3.50687, p- value = 0.0001), whereas the association was lost for S. musiva leaf spot and a new association was revealed on Chr06 (marker position = 142.6761866, p-value = 0.017). The top Chr04 marker explained 57.9% of variation in Melampsora sp. infection severity while the new Chr06 association explained 4.77% of the variation in S. musiva leaf spot disease severity. The positive allele contributing to an increase in fungal infection on Chr04 was derived from P. deltoides whereas the positive allele contributing to severity of S. musiva leaf spot infection on Chr06 was derived from

P. trichocarpa. Binary presence of the canker symptoms caused by S. musiva (Figure 5a) was found to be associated with a QTL located on Chr16 (marker position = 60.81429, p-value =

0.022). The positive allele contributing to increase in presence of S. musiva canker was from P. trichocarpa.

Three QTL intervals containing a total of 40 markers were significantly associated with insect abundance (Table 2, Figure 5b,c). A QTL on Chr05 (marker position = 21.83874, p-value = 0.001) was associated with binary presence of M. vagabunda branch galls (Figure 5b). The positive allele contributing to an increase in presence of M. vagabunda was from P. trichocarpa. Two QTLs, one located on Chr10 (marker position = 57.69914343, p-value = 0.015) and one located on Chr13

(marker position = 84.43857329, p-value = 0.045), were associated with the number of leaf-folding galls from Phyllocolpa sp. (Figure 5c). The marker with the highest LOD score explained 9.65% of the variance in oviposition gall count while 8.82% of the variance was explained by the top marker located on Chr13. For both markers, the positive allele contributing to an increase in the number of female oviposition galls was from P. deltoides. There were no QTL intervals that passed the permutation threshold for the binary presence of P. populitransversus petiole galls (Figure 5c).

Interaction between Melampsora sp. and S. musiva leaf infection

Melampsora sp. leaf rust infection severity was dependent upon the disease severity of S. musiva leaf spot symptoms (F = 35.2, p-value = 0.0001) (Table 3). The severity of Melampsora sp. infection for individuals with no S. musiva leaf spot was significantly higher than for individuals with S. musiva leaf spot symptoms (Figure 6). Similarly, S. musiva leaf spot disease severity was dependent upon the levels of Melampsora sp. leaf rust (F = 31.5, p-value = 0.0001). S. musiva leaf spot infection severity was significantly lower for individuals with a Melampsora sp. leaf spot score of 3 compared to individuals with less severe Melampsora sp. leaf spot symptoms (Figure 6).

PFAM, GO term and tandem duplication enrichment

The total number of genes present in the QTL intervals of each parental species was 174 on Chr04,

81 on Chr16, 161 on Chr10, 445 on Chr13, and 156 on Chr05 (Table 4). There were 40 PFAM domains enriched relative to the rest of the P. trichocarpa genome across all QTL intervals (alpha

= 0.05; p-value threshold < 0.00014), and 25 PFAM domains enriched in the P. deltoides QTL intervals (alpha = 0.05, p-value < 0.000059). Additionally, there were 21 GO terms enriched in the

P. trichocarpa intervals (alpha = 0.05, p-value < 0.000226) and 46 GO terms enriched in the P. deltoides intervals (alpha = 0.05, p-value < 0.000214). Finally, seven of the genes in these intervals showed evidence of positive selection based on the ratio of nonsynonymous to synonymous nucleotide substitutions between the P. deltoides and P. trichocarpa orthologs (Table 5).

In total, there were 107 recent tandem duplicates in the P. trichocarpa intervals and 195 in the P. deltoides intervals for all biotic QTLs (Figure 7, Table 6). This includes 23 and 41 recent tandem duplicates for the Melampsora sp. intervals; 6 and 10 for the S. musiva intervals; 32 and 96 for the

Phyllocolpa sp. intervals; and 47 and 46 for the M. vagabunda intervals, for P. trichocarpa and P. deltoides, respectively. The total number of recent tandem duplicates was significantly enriched relative to tandem counts for random intervals of the same size as the QTL intervals for the P. deltoides grandparent (p-value = 0.0118), but not for the P. trichocarpa grandparent (p-value =

0.1191).

2.6 Discussion

The goal of our research was to utilize QTL analysis as a tool to identify regions of the Populus genome that were important in mediating biotic interactions. Upon identification of these regions, we were able to directly compare the parental genomes of the hybrid cross to look for similarity in

content and potential gene-for-gene interactions reflected in recent tandem duplication expansion.

We found that the host plant genotype had a significant effect on fungi and insects in our study.

Additionally, the progeny segregated for varying resistance to fungal and insect pathogens and pests that were inherited from the grandparents and parents of the 52-124 family cross.

Novel alleles (i.e. those from the grandparent that were not native to the region where the trials occurred) appear to be important for resistance to Melampsora sp. fungus, M. vagabunda galls, and

Phyllocolpa sp. oviposition. This is not the first case of potential novel alleles being important in resistance to biotic stress. In a P. trichocarpa GWAS population, three genes have been found that confer novel resistance to S. musiva canker infection and an additional single gene that, when inherited, can suppress that resistance (Muchero et al., 2018). Although cases such as this are still under investigation to understand mechanisms of novel resistance can also originate from a variety of simple traits inherent to a non-native species, such as delayed emergence due to phenology

(Mercader et al., 2009) or changes in secondary metabolites that are important cues in insect recognition of the host important in oviposition (Nahrstedt, 1989). In contrast, the interval associated with S. musiva canker symptoms was the only case in which susceptibility to the fungus was dominant and inherited from the non-coevolved host (P. trichocarpa), as has been previously observed (Newcombe & Ostry, 2001; Muchero et al., 2018).

We detected a major QTL for Melampsora sp. resistance on Chr04. Although the specific strain that infected the trees is unknown, this genomic interval is known to contain the MXC3 locus, which confers resistance to infection of multiple species of the Melampsora leaf rusts (Newcombe et al., 2001; Yin et al., 2004a). Based on mean parental infection scores and the allelic effects at the

QTL, progeny in family 52-124 inherited this resistance from the P. trichocarpa grandmother 93-

968. Furthermore, the QTL interval contained two tandem repeats of stigma-specific proteins

(Stig1) in both the P. trichocarpa genome and the orthologous interval in the P. deltoides genome.

Several of these genes showed evidence of positive selection based on Ka/Ks ratios (Table 5).

Stigma-specific proteins, specifically Stig1, have been found to be associated with female sterility in tobacco (Nicotiana tabacum) and petunia (Petunia hybrida) (Goldman et al., 1994; Verhoeven et al., 2005). Stig1 is known to mediate secretion of exudate lipids in the intercellular spaces and high expression of the protein inhibits pollen grains from penetrating style tissue preventing fertilization

(Verhoeven et al., 2005). Most lipid transfer proteins, such as Stig1, are important in plant cell-wall loosening and their expression can prevent penetration of plant tissues (Nieuwland et al., 2005).

Diversification of the protein family containing Stig1 may play an important role in lowering fungal infection in Populus by providing a physical barrier to resist the Melampsora sp. hyphae.

Another protein domain that was found to be enriched in the P. trichocarpa Chr04 interval was the malonyl-CoA decarboxylase C-terminal domain. Similarly, the gene ontology function for malonyl-CoA decarboxylase activity was also enriched in P. trichocarpa in the same interval.

Genes products that are capable of transforming malonyl-CoA are important precursors in the production of several pathogen defensive compounds, such as isoprenoids in the mevalonate pathway (Dixon, 2001; Chen et al., 2011).

Interestingly, an overlapping interval on Chr04 was found to be associated with the activity of the leaf spot symptoms of the S. musiva fungus. Upon further investigation, we found that trees that were not infected by the S. musiva leaf spot had more severe symptoms of the Melampsora sp. leaf rust. This suggests that a competitive interaction occurred between the two pathogens in the field during the year of survey and was reflected in an inflated association on Chr04 with the S. musiva leaf spot score. Furthermore, when individuals with presence of Melampsora sp. infection were

removed from the S. musiva analysis the association with Chr04 is no longer significant, and a new

QTL appeared on Chr06. Competition among fungal pathogens is not uncommon in field conditions, with most examples focused on different genotypes within the same fungal species

(Abdullah et al., 2017). The outcome of within host tissue colonization of multiple strains or species often relies upon the genetic similarity of the pathogens (Koskella et al., 2006; Abdullah et al., 2017). In our case, we had two very different pathogens that were utilizing the leaf tissue in vastly dissimilar ways. S. musiva is a necrotrophic fungus, which requires dead tissue to reproduce, and Melampsora sp. is biotrophic, requiring living tissue to reproduce. The presence of S. musiva on the trees had a much larger effect on the occurrence of Melampsora sp. infection. This may indicate that the S. musiva fungus had a competitive advantage over Melampsora sp. in utilization of the Populus leaf tissue at our site. Further supporting this finding was the complete loss of

Melampsora sp. symptoms at the plantation site over the course of ten years and the continual presence of the S. musiva leaf spot. A similar interaction has been reported in wheat between the biotroph Blumeria graminis f.sp. tritici (Bgt), the powdery mildew pathogen, and necrotroph

Zymoseptoria tritici, the cause of Septoria tritici blotch (Orton & Brown, 2016). Similarly, the outcome of the interaction of the two pathogens in wheat was competitive. However, the necrotroph was found to actually be capable of reducing the reproductive capability of the biotroph, which indicates pathogen-pathogen interactions can be more direct rather than relying solely on the host plant genetics (Orton & Brown, 2016).

The S. musiva leaf spot and stem canker symptoms were found to be associated with QTL intervals on Chr06 and Chr16, respectively. In this study, susceptibility to the necrotrophic fungi appeared to be originating from the presence of P. trichocarpa alleles in the progeny for both symptoms.

Previous work on a similar Populus hybrid crosses support this, with susceptibility to S. musiva

necrotrophic fungi originating primarily from dominant alleles derived from P. trichocarpa

(Newcombe, 1998; Newcombe & Ostry, 2001; Muchero et al., 2018). The Chr06 leaf spot association contained numerous genes with methyltransferase activity, which were higher in the P. deltoides interval. Methyltransferase enzymes are important in plant secondary metabolism and have been found to be key in the production of a variety of anti-microbial compounds (Noel et al.,

2003). Interestingly, the loci conferring susceptibility in family 52-124 in the Chr16 canker interval did not overlap with the four loci that were uncovered in a previous genome-wide association study of S. musiva susceptibility in P. trichocarpa (Muchero et al., 2018), suggesting that different mechanisms may be involved in hybrid interactions with this pathogen. However, the QTL did contain a tandem repeat of a G-type lectin receptor-like protein kinase that was expanded in P. deltoides. This protein could play a similar role to a receptor-like kinase from the same family

(Yang et al., 2016) that was associated with susceptibility to S. musiva in the P. trichocarpa study

(Muchero et al., 2018).

M. vagabunda has been recorded completing its life cycle on several species of Populus, including

P. deltoides and P. tremuloides (Floate, 2010). Although the aphid’s life history has been documented, little is known about the influence of host plant genetics on gall formation or resistance to feeding (Ignoffo & Granovsky, 1961b; Floate, 2010). We detected a QTL on Chr05 in which alleles inherited from P. deltoides were positively associated with gall occurrence. Genes conferring lipoxygenase and oxidoreductase activity were enriched in both the P. deltoides as well as the P. trichocarpa QTL intervals. Lipoxygenase genes are known to be associated with the

Populus response to both abiotic and biotic stressors (Ralph et al., 2006; Cheng et al., 2006). They are often upregulated in the presence of mechanical damage, fungal pathogen invasion, and exposure to simulated insect feeding (Cheng et al., 2006; Chen et al., 2009). The lipoxygenases are

important in the formation of jasmonic acid, the signaling molecule that upregulates plant defenses against herbivore feeding (Chen et al., 2009).

The M. vagabunda QTL interval on Chr05 also contained a tandem array of resistance genes (R- genes) that encoded disease resistance proteins (TIR-NBS-LRR class) that was greatly expanded in

P. trichocarpa compared to P. deltoides, as well as repeats of the leucine-rich repeat protein kinase family proteins in both species. These protein families are well known for their roles in the recognition and upregulation of host plant defenses against bacterial and fungal infection

(Bergelson et al., 2001; Martin et al., 2003). Neofunctionalization of R-genes through tandem duplication due to gene-for-gene coevolution has also been demonstrated in many plant-fungal pathosystems (Leister, 2004). Although R-genes have been more frequently related to plant-fungal interactions, they are also becoming commonly associated in mediating plant-insect interactions as well, especially in insects that utilize piercing-sucking feeding (Harris et al., 2003; Kaloshian,

2004).

In addition to the R-genes, there were a series of genes encoding cytochrome P450 family proteins in both species as well as a unique tandem set that was only present in the P. trichocarpa genome.

P450 enzymes are important in the production of many classes of secondary metabolites, such as furanocoumarins and terpenoids, which are highly toxic to insects (Keeling & Bohlmann, 2006;

Schuler, 2011). Alternatively, cytochrome P450’s may also be implicated in the susceptibility of the host plant to galling aphids. They are important in the synthesis of fatty acids and production of suberin in plant tissues (Höfer et al., 2008; Pinot & Beisson, 2011). Typically, suberin is important in separation of different tissues as well as in the establishment of apoplastic barriers that restrict nutrient/water loss as well as pathogen invasion (Höfer et al., 2008; Qin & LeBoldus, 2014).

Several insects are known to produce suberized spherical galls on leaves, including hymenopteran pests of Rosacea species and dipteran pests of Fabaceae (Krishnan & Franceschi, 1988; Oliveira et al., 2016). If aphids induce the suberization mechanism of the plant genome, may lead to the increased toughening of aphid galls, much like the woody structures M. vagabunda leaves behind on branches once they have finished feeding.

We detected two QTL for Phyllocolpa sp., with P. trichocarpa alleles being positively associated with leaf fold occurrence in both cases. The P. trichocarpa genomic interval corresponding to the

Chr10 QTL was enriched for several domains and GO terms that could be involved in gall development. For example, the interval was enriched for sugar transporters as well as GO terms for sucrose transporter activity. Often, in the case of galling insects, plant tissue is modified in such a way as to act as a sugar sink, thereby enhancing its nutritional value to larvae (Larson & Whitham,

1991; Nyman et al., 2000; Wool, 2004) . The presence of these combinations of sugar transporter genes may be mediating a similar interaction between the Phyllocolpa sp. female sawflies and their chosen Populus hosts.

Phyllocolpa sp. galls are formed early in the season when a female sawfly selects a leaf and injects the longitudinal fold with small amounts of fluid on the underside of young leaves (Fritz & Price,

1988; Kopelke, 2007). The adult sawfly will proceed to oviposit near the base of the leaf, and after

1-2 days, the leaf fold forms and the newly hatched larvae feeds on the inside of the gall (Smith &

Fritz, 1996). The Phyllocolpa sp. galls were a unique biotic phenotype to this study as they were an estimate of female sawfly ovipositional choice rather than feeding success. Host selection for oviposition is initially driven by visual cues and reinforced by females assessing the nutrition and chemical cues of foliage (Panda & Khush, 1995; Boeckler et al., 2011). Previous research in Salix

(closely related to Populus) has shown that common phenolic glycosides in leaf tissue are important in host choice in both the free-feeding Nematus oligospilus and galling Euura amerinae specialist sawflies (Kolehmainen et al., 1994; Fernández et al., 2019). We therefore pursued QTL mapping of metabolites to determine if phenolic compounds might be important determinants of the potential ovipositional relationship.

We detected three significant QTL for gentisyl alcohol 5-O-glucoside levels. One of these overlapped with the Phyllocolpa sp. leaf fold QTL on Chr10. The QTL on Chr14 also overlapped with a suggestive QTL peak for Phyllocolpa sp. (LOD=3.96). This is the first case of an association of gentisyl alcohol 5-O-glucoside with a specialist arthropod in Populus, and is, in fact, the first report of this metabolite in Populus. Salirepin (gentisyl alcohol 2-O-glucoside), a closely related metabolite, is a well-known constituent in Populus spp., (Busov et al., 2006; Veach et al.,

2018; Tschaplinski et al., 2019) and leaves of P. deltoides and P. trichocarpa x deltoides have a lower abundance, later eluting metabolite with a nearly identical fragmentation pattern to salirepin that we tentatively identify as gentisyl alcohol 2-O-glucoside. The QTL contains three putative candidate genes, including an aldehyde dehydrogenase 5F1 (Podel.10G175800) that may be involved in the reduction of gentisyl aldehyde to the alcohol, and two UDP-glycosyltransferases

(Podel.10G184800, Podel.10G185000) that may be involved in the gentisyl alcohol conjugation to glucose. Specific substrates have yet to be determined for these genes, but a previous report suggests aldehyde dehydrogenase 5F1 genes are likely involved in the basic metabolism of Populus (Tian et al., 2015).

Phyllocolpa sp. sawflies are considered a keystone species as the abandoned or unused leaf folds are often used as a habitat for many other species, such as aphids and spiders (Bailey & Whitham,

2007). The presence of folds in aspen forests is associated with a two-fold increase in arthropod species richness and around a four-fold increase in arthropod abundance relative to forests where the insect is absent (Bailey & Whitham, 2003). This in turn makes the host plant and sawfly relationship important in examining how shifts in the genes of a population ultimately structure whole communities, effectively linking ecology and evolutionary biology. Further investigation of this potential relationship could be key to connecting Populus genetics to the assemblage of the surrounding communities of organisms.

A striking finding in this study was an elevated number of recent tandem duplications in the P. deltoides genome, but not the P. trichocarpa genome, for the biotic QTL intervals. Out of the six chromosomes that yielded significant QTL results, four were associated with phenotypes that were fungi and insects native to the distribution of P. deltoides, but not P. trichocarpa. Given that P. deltoides has been co-evolving with the majority of the surveyed fungi and insects, it was not unexpected that there were more recent tandem duplicates in biotic intervals in its genome as there is more selective pressure on the native species to overcome biotic stress (Constabel & Lindroth,

2010; Newcombe et al., 2010). However, given the high amount of novel resistance occurring in the progeny, recent tandem duplication may also be important in naïve host resistance.

In our study, we demonstrated how host plant genetics directly affect associated fungi and insects in the field, as well as how Populus progeny indirectly structured interactions between pathogens

(Whitham et al., 2006). The competition between Melampsora sp. and S. musiva highlights the complexities of how hybrid genetics are capable of strongly mediating multiple species interactions, which can result in inflated genetic associations. Finally, we have shown that many recent tandem duplications, found across biotic stress QTL intervals, have functional annotations

that are involved in host plant physical/chemical resistance and tolerance as well as a few that may be implicated in host plant susceptibility. The enrichment of recent tandem duplications is a signature of gene-for-gene interactions, and a mechanism that is essential to protect long-lived plants such as trees, enabling them to reach maturity despite many coevolving biotic stressors.

2.7 Literature cited

Abdullah AS, Moffat CS, Lopez-Ruiz FJ, Gibberd MR, Hamblin J, Zerihun A. 2017. Host–

multi-pathogen warfare: pathogen interactions in co-infected plants. Frontiers in Plant

Science 8: 1806.

Bailey JK, Whitham TG. 2003. Interactions among elk, aspen, galling sawflies and insectivorous

birds. Oikos 101: 127–134.

Bailey JK, Whitham TG. 2007. Biodiversity is related to indirect interactions among species of

large effect. In: Ecological Communities: Plant Mediation in Indirect Interaction Webs.

Cambridge: Cambridge University Press, 306–328.

Bergelson J, Kreitman M, Stahl EA, Tian D. 2001. Evolutionary dynamics of plant R-genes.

Science 292: 2281–2285.

Bird J, Faith DP, Rhomberg L, Riska B, Sokal RR. 1979. The morphs of Pemphigus

populitransversus: allocation methods, morphometrics, and distribution patterns. Annals of

the Entomological Society of America 72: 767–774.

Boeckler GA, Gershenzon J, Unsicker SB. 2011. Phenolic glycosides of the Salicaceae and their

role as anti-herbivore defenses. Phytochemistry 72: 1497–509.

Broman KW, Wu H, Sen S, Churchill GA. 2003. R/qtl: QTL mapping in experimental crosses.

Bioinformatics 19: 889–890.

Busov V, Meilan R, Pearce DW, Rood SB, Ma C, Tschaplinski TJ, Strauss SH. 2006.

Transgenic modification of gai or rgl1 causes dwarfing and alters gibberellins, root growth,

and metabolite profiles in Populus. Planta 224: 288–299.

Chen XM. 2005. Epidemiology and control of stripe rust [Puccinia striiformis f. sp. tritici] on

wheat. Canadian Journal of Plant Pathology 27: 314–337.

Chen H, Kim HU, Weng H, Browse J. 2011. Malonyl-CoA synthetase, encoded by ACYL

ACTIVATING ENZYME13, is essential for growth and development of Arabidopsis. The

Plant Cell 23: 2247–62.

Chen F, Liu C-J, Tschaplinski TJ, Zhao N. 2009. Genomics of secondary metabolism in

Populus: interactions with biotic and abiotic environments. Critical Reviews in Plant

Sciences 28: 375–392.

Cheng Q, Zhang B, Zhuge Q, Zeng Y, Wang M, Huang M. 2006. Expression profiles of two

novel lipoxygenase genes in Populus deltoides. Plant Science 170: 1027–1035.

Chisholm ST, Coaker G, Day B, Staskawicz BJ. 2006. Host-microbe interactions: shaping the

evolution of the plant immune response. Cell 124: 803–814.

Constabel CP, Lindroth RL. 2010. The impact of genomics on advances in herbivore defense and

secondary metabolism in Populus. In: Jansson S, Bhalerao R, Groover A, eds. Genetics and

Genomics of Populus. NY: Springer New York, 279–305.

Crutsinger GM, Rudman SM, Rodriguez-Cabal MA, McKown AD, Sato T, MacDonald AM,

Heavyside J, Geraldes A, Hart EM, LeRoy CJ, et al. 2014. Testing a ‘genes-to-

ecosystems’ approach to understanding aquatic-terrestrial linkages. Molecular Ecology 23:

5888–5903.

Danecek P, Auton A, Abecasis G, Albers CA, Banks E, DePristo MA, Handsaker RE, Lunter

G, Marth GT, Sherry ST, et al. 2011. The variant call format and VCFtools.

Bioinformatics 27: 2156–2158.

Dixon RA. 2001. Natural products and plant disease resistance. Nature 411: 843–847.

Ehrlich PR, Raven PH. 1964. Butterflies and plants: a study in coevolution. Evolution 18: 586–

608.

Faith DP. 1979. Strategies of gall formation in Pemphigus aphids. Journal of the New York

Entomological Society 87: 21–37.

Fernández PC, Braccini CL, Dávila C, Barrozo RB, Aráoz MVC, Cerrillo T, Gershenzon J,

Reichelt M, Zavala JA. 2019. The use of leaf surface contact cues during oviposition

explains field preferences in the willow sawfly Nematus oligospilus. Scientific Reports 9:

4946.

Floate KD. 2010. Gall-inducing aphids and mites associated with the hybrid complex of

cottonwoods, Populus spp. (Salicaceae) on Canada’s grasslands. Arthropods of Canadian

Grasslands 1: 281–300.

Flor HH. 1971. Current status of the gene-for-gene concept. Annual Review of Phytopathology 9:

275–296.

Friesen TL, Meinhardt SW, Faris JD. 2007. The Stagonospora nodorum-wheat pathosystem

involves multiple proteinaceous host-selective toxins and corresponding host sensitivity

genes that interact in an inverse gene-for-gene manner. Plant Journal 51: 681–692.

Fritz RS, Price PW. 1988. Genetic variation among plants and insect community structure:

willows and sawflies. Ecology 69: 845–856.

Gallun RL. 1977. Genetic basis of Hessian fly epidemics. Annals of the New York Academy of

Sciences 287: 223–229.

Gianola D, Norton HW. 1981. Scaling threshold characters. Genetics 99: 357–364.

Goldman MH, Goldberg RB, Mariani C. 1994. Female sterile tobacco plants are produced by

stigma‐specific cell ablation. The EMBO Journal 13: 2976–2984.

Goodstein DM, Shu S, Howson R, Neupane R, Hayes RD, Fazo J, Mitros T, Dirks W, Hellsten

U, Putnam N, et al. 2012. Phytozome: a comparative platform for green plant genomics.

Nucleic Acids Research 40: D1178–D1186.

Harris MO, Stuart JJ, Mohan M, Nair S, Lamb RJ, Rohfritsch O. 2003. Grasses and gall

midges: plant defense and insect adaptation. Annual Review of Entomology 48: 549–577.

Haruta M, Major IT, Christopher ME, Patton JJ, Constabel CP. 2001. A Kunitz trypsin

inhibitor gene family from trembling aspen (Populus tremuloides Michx.): cloning,

functional expression, and induction by wounding and herbivory. Plant Molecular Biology

46: 347–359.

Höfer R, Briesen I, Beck M, Pinot F, Schreiber L, Franke R. 2008. The Arabidopsis

cytochrome P450 CYP86A1 encodes a fatty acid ω-hydroxylase involved in suberin

monomer biosynthesis. Journal of Experimental Botany 59: 2347–2360.

Hulbert SH, Webb CA, Smith SM, Sun Q. 2001. Resistance gene complexes: evolution and

utilization. Annual Review of Phytopathology 39: 285–312.

Ignoffo CM, Granovsky AA. 1961a. Life history and gall development of Mordwilkoja

vagabunda (Homoptera: Aphidae) on Populus deltoides. Part II—gall development. Annals

of the Entomological Society of America 54: 635–641.

Ignoffo CM, Granovsky AA. 1961b. Life history and gall development of Mordwilkoja

vagabunda (Homoptera: Aphidae) on Populus deltoides. Annals of the Entomological

Society of America 54: 486–499.

Kaloshian I. 2004. Gene-for-gene disease resistance: bridging insect pest and pathogen defense.

Journal of Chemical Ecology 30: 2419–2438.

Keeling CI, Bohlmann J. 2006. Genes, enzymes and chemicals of terpenoid diversity in the

constitutive and induced defence of conifers against insects and pathogens. New Phytologist

170: 657–675.

Kolehmainen J, Roininen H, Julkunen-Tiitto R, Tahvanainen J. 1994. Importance of phenolic

glucosides in host selection of shoot galling sawfly, Euura amerinae, on Salix pentandra.

Journal of Chemical Ecology 20: 2455–2466.

Kopelke J-P. 2007. The European species of the genus Phyllocolpa, Part I: The leucosticta-group

(Insecta, Hymenoptera, Tenthredinidae, Nematinae). Senckenbergiana Biologica 87: 75–

109.

Koskella B, Giraud T, Hood ME. 2006. Pathogen relatedness affects the prevalence of within-

host competition. The American Naturalist 168: 121–126.

Krishnan HB, Franceschi VR. 1988. Anatomy of some leaf galls of Rosa woodsii (Rosaceae).

American Journal of Botany 75: 396–376.

Larson KC, Whitham TG. 1991. Manipulation of food resources by a gall-forming aphid: the

physiology of sink-source interactions. Oecologia 88: 15–21.

Lefebvre V, Chèvre A. 1995. Tools for marking plant disease and pest resistance genes: a review.

Agronomie 15: 3–19.

Leister D. 2004. Tandem and segmental gene duplication and recombination in the evolution of

plant disease resistance genes. Trends in Genetics 20: 116–122.

Li H. 2011. A statistical framework for SNP calling, mutation discovery, association mapping and

population genetical parameter estimation from sequencing data. Bioinformatics 27: 2987–

2993.

Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G, Abecasis G, Durbin

R. 2009. The Sequence Alignment/Map format and SAMtools. Bioinformatics 25: 2078–

2079.

Major IT, Constabel CP. 2008. Functional analysis of the Kunitz trypsin inhibitor family in

poplar reveals biochemical diversity and multiplicity in defense against herbivores. Plant

Physiology 146: 888–903.

Martin GB, Bogdanove AJ, Sessa G. 2003. Understanding the functions of plant disease

resistance proteins. Annual Review of Plant Biology 54: 23–61.

McKown AD, Klapste J, Guy RD, Geraldes A, Porth I, Hannemann J, Friedmann M,

Muchero W, Tuskan GA, Ehlting J, et al. 2014. Genome-wide association implicates

numerous genes underlying ecological trait variation in natural populations of Populus

trichocarpa. New Phytologist 203: 535–553.

Meilan R, Han KH, Ma C, DiFazio SP, Eaton JA, Hoien EA, Stanton BJ, Crockett RP, Taylor

ML, James RR, et al. 2002. The CP4 transgene provides high levels of tolerance to

Roundup® herbicide in field-grown hybrid poplars. Canadian Journal of Forest Research

32: 967–976.

Mello MO, Silva-Filho MC. 2002. Plant-insect interactions: an evolutionary arms race between

two distinct defense mechanisms. Brazilian Journal of Plant Physiology 14: 71–81.

Mercader RJ, Aardema ML, Scriber JM. 2009. Hybridization leads to host-use divergence in a

polyphagous butterfly sibling species pair. Oecologia 158: 651–662.

Muchero W, Guo J, DiFazio SP, Chen J-G, Ranjan P, Slavov GT, Gunter LE, Jawdy S,

Bryan AC, Sykes R, et al. 2015. High-resolution genetic mapping of allelic variants

associated with cell wall chemistry in Populus. BMC Genomics 16: 24.

Muchero W, Sondreli KL, Chen JG, Urbanowicz BR, Zhang J, Singan V, Yang Y,

Brueggeman RS, Franco-Coronado J, Abraham N, et al. 2018. Association mapping,

transcriptomics, and transient expression identify candidate genes mediating plant-pathogen

interactions in a tree. Proceedings of the National Academy of Sciences of the United States

of America 115: 11573–11578.

Nahrstedt A. 1989. The significance of secondary metabolites for interactions between plants and

insects. Planta Medica 55: 333–338.

Newcombe G. 1998. A review of exapted resistance to diseases of Populus. Forest Pathology 28:

209–216.

Newcombe G, Martin F, Kohler A. 2010. Defense and nutrient mutualisms in Populus. In:

Jansson S, Bhalerao R, Groover A, eds. Genetics and Genomics of Populus. NY: Springer

New York, 247–277.

Newcombe G, Ostry M. 2001. Recessive resistance to Septoria stem canker of hybrid poplar.

Phytopathology 91: 1081–1084.

Newcombe G, Stirling B, Bradshaw Jr. HD. 2001. Abundant pathogenic variation in the new

hybrid rust Melampsora × columbiana on hybrid poplar. Phytopathology 91: 981–985.

Nieuwland J, Feron R, Huisman BAH, Fasolino A, Hilbers CW, Derksen J, Mariani C. 2005.

Lipid transfer proteins enhance cell wall extension in tobacco. The Plant Cell Online 17:

2009–2019.

Noel JP, Dixon RA, Pichersky E, Zubieta C, Ferrer JL. 2003. Structural, functional, and

evolutionary basis for methylation of plant small molecules. In: Romeo J, ed. Recent

Advances in Phytochemistry. Oxford: Elsevier Science, 37–58.

Nyman T, Widmer A, Roininen H. 2000. Evolution of gall morphology and host-plant

relationships in willow-feeding sawflies (Hymenoptera: Tenthradinidae). Evolution 54:

526–533.

Oliveira DC, Isaias RM, Fernandes GW, Ferreira BG, Carneiro RG, Fuzaro L. 2016.

Manipulation of host plant cells and tissues by gall-inducing insects and adaptive strategies

used by different feeding guilds. Journal of Insect Physiology 1: 103–113.

Orton ES, Brown JKM. 2016. Reduction of growth and reproduction of the biotrophic fungus

Blumeria graminis in the presence of a necrotrophic pathogen. Frontiers in Plant Science 7:

742.

Panda N, Khush GS. 1995. Host plant resistance to insects. Wallingford, UK: CAB International.

Philippe RN, Ralph SG, Külheim C, Jancsik SI, Bohlmann J. 2009. Poplar defense against

insects: genome analysis, full-length cDNA cloning, and transcriptome and protein analysis

of the poplar Kunitz-type protease inhibitor family. New Phytologist 184: 865–884.

Pinot F, Beisson F. 2011. Cytochrome P450 metabolizing fatty acids in plants: characterization

and physiological roles. The FEBS journal 278: 195–205.

Pretorius ZA, Singh RP, Wagoire WW, Payne TS. 2000. Detection of virulence to wheat stem

rust resistance gene Sr31 in Puccinia graminis. f. sp. tritici in Uganda. Plant Disease 84:

203–203.

Qin R, LeBoldus JM. 2014. The infection biology of Sphaerulina musiva: clues to understanding

a forest pathogen. PLoS ONE 9: e103477.

Ralph S, Oddy C, Cooper D, Yueh H, Jancsik S, Kolosova N, Philippe RN, Aeschliman D,

White R, Huber D, et al. 2006. Genomics of hybrid poplar (Populus trichocarpa ×

deltoides) interacting with forest tent caterpillars (Malacosoma disstria): normalized and

full-length cDNA libraries, expressed sequence tags, and a cDNA microarray for the study

of in. Molecular Ecology 15: 1275–1297.

Schuler MA. 2011. P450s in plant–insect interactions. Biochimica et Biophysica Acta (BBA) -

Proteins and Proteomics 1814: 36–45.

Slavov G, DiFazio S, Martin J, Schackwitz W, Muchero W, Rodgers-Melnick E, Lipphardt

M, Pennacchio C, Hellsten U, Pennacchio L, et al. 2012. Genome resequencing reveals

multiscale geographic structure and extensive linkage disequilibrium in the forest tree

Populus trichocarpa. New Phytologist 196: 713–725.

Smith DR, Fritz RS. 1996. Review of the eastern United States species of the leaf-folding sawflies

of the genus Phyllocolpa Benson (Hymenoptera: Tenthredinidae). Proceedings of the

Entomological Society of Washington 98: 695–707.

Stajich JE. 2002. The Bioperl Toolkit: perl modules for the life sciences. Genome Research 12:

1611–1618.

Stanton BJ, B. Neale D, Li S. 2010. Populus breeding: from the classical to the genomic approach.

In: Jansson S, Bhalerao R, Groover A, eds. Genetics and Genomics of Populus. Springer

New York, 309–348.

Taylor G. 2002. Populus: Arabidopsis for forestry. Do we need a model tree? Annals of Botany 90:

681–689.

Thompson JN. 1988. Coevolution and alternative hypotheses on insect/plant interactions. Ecology

69: 893–895.

Thompson JN, Burdon JJ. 1992. Gene-for-gene coevolution between plants and parasites. Nature

360: 121–125.

Tian F-X, Zang J-L, Wang T, Xie Y-L, Zhang J, Hu J-J. 2015. Aldehyde dehydrogenase gene

superfamily in Populus: organization and expression divergence between paralogous gene

pairs (A Palsson, Ed.). PLoS ONE 10: e0124669.

Tschaplinski TJ, Abraham PE, Jawdy SS, Gunter LE, Martin MZ, Engle NL, Yang X,

Tuskan GA. 2019. The nature of the progression of drought stress drives differential

metabolomic responses in Populus deltoides. Annals of Botany 124: 617–626.

Veach AM, Yip D, Engle NL, Yang ZK, Bible A, Morrell-Falvey J, Tschaplinski TJ, Kalluri

UC, Schadt CW. 2018. Modification of plant cell wall chemistry impacts metabolome and

microbiome composition in Populus PdKOR1 RNAi plants. Plant and Soil 429: 349–361.

Verhoeven T, Feron R, Wolters-Arts M, Edqvist J, Gerats T, Derksen J, Mariani C. 2005.

STIG1 controls exudate secretion in the pistil of petunia and tobacco. Plant Physiology 138:

153–160.

Verkley GJM, Quaedvlieg W, Shin H-D, Crous PW. 2013. A new approach to species

delimitation in Septoria. Studies in Mycology 75: 213–305.

Wang Y, Tang H, Debarry JD, Tan X, Li J, Wang X, Lee TH, Jin H, Marler B, Guo H, et al.

2012. MCScanX: A toolkit for detection and evolutionary analysis of gene synteny and

collinearity. Nucleic Acids Research 40: 1–14.

Weston DJ, Rogers A, Tschaplinski TJ, Gunter LE, Jawdy SA, Engle NL, Heady LE, Tuskan

GA, Wullschleger SD. 2015. Scaling nitrogen and carbon interactions: what are the

consequences of biological buffering? Ecology and Evolution 5: 2839–2850.

Whitham TG, Bailey JK, Schweitzer JA, Shuster SM, Bangert RK, LeRoy CJ, Lonsdorf E V.,

Allan GJ, DiFazio SP, Potts BM, et al. 2006. A framework for community and ecosystem

genetics: from genes to ecosystems. Nature Reviews Genetics 7: 510–523.

Wool D. 2004. Galling aphids: specialization, biological complexity, and variation. Annual Review

of Entomology 49: 175–192.

Yang Y, Labbé J, Muchero W, Yang X, Jawdy SS, Kennedy M, Johnson J, Sreedasyam A,

Schmutz J, Tuskan GA, et al. 2016. Genome-wide analysis of lectin receptor-like kinases

in Populus. BMC Genomics 17: 699.

Yin T-M, DiFazio SP, Gunter LE, Jawdy SS, Boerjan W, Tuskan GA. 2004a. Genetic and

physical mapping of Melampsora rust resistance genes in Populus and characterization of

linkage disequilibrium and flanking genomic sequence. New Phytologist 164: 95–105.

Yin T-M, DiFazio SP, Gunter LE, Riemenschneider D, Tuskan GA. 2004b. Large-scale

heterospecific segregation distortion in Populus revealed by a dense genetic map.

Theoretical and Applied Genetics 109: 451–463.

Zhao N, Guan J, Ferrer JL, Engle N, Chern M, Ronald P, Tschaplinski TJ, Chen F. 2010.

Biosynthesis and emission of insect-induced methyl salicylate and methyl benzoate from

rice. Plant Physiology and Biochemistry 48: 279–287.

2.8 Tables and Figures

Table 1. TPS and linear-mixed model output for biotic surveys. Broad-sense heritability (H2) denotes the contribution of all host plant genetic factors to total variance in the biotic phenotype.

R package RLRsim exactRLRT function was used to test significance of effects in mixed-model.

Fungus/Insect TPS R2 TPS H2 Genetic Error RLRatio p-value RMSE variance variance Melampsora sp. 18.96% 1.037 0.609 0.271 0.174 294 < 0.0001 leaf rust

S. musiva 15.33% 0.6757 0.250 0.153 0.459 32.0 < 0.0001 leaf spot

Phyllocolpa sp. 4.537% 5.469 0.391 11.9 18.5 129 < 0.0001

Table 2. Summary of QTL permutation test output. Percent variance in surveys for insects and fungi explained by significant marker indicated for composite interval mapping models. Positive (+) allele specifies genotype at significant interval that results in an increase in susceptibility. D indicates progeny are homozygous for P. deltoides alleles and T indicates progeny are heterozygous for P. deltoides and P. trichocarpa alleles.

Model Chrom. LOD score p-value % variance + allele Garden

Melampsora sp. leaf rust Chr04 19.7 0.001 52.5 D Morgantown all individuals

Melampsora sp. leaf rust Chr04 30.1 <0.0001 57.9 D Morgantown subsetted individuals S. musiva leaf spot all individuals Chr04 7.80 0.001 5.29 D Morgantown

S. musiva leaf spot Chr06 4.60 <0.0001 4.77 T Morgantown subsetted individuals S. musiva canker Chr16 4.34 0.022 NA T Morgantown M. vagabunda Chr05 5.39 0.001 NA D Morgantown P. populitransversus Chr03 2.70 >0.05 NA NA Morgantown Phyllocolpa sp. Chr10 6.60 0.015 9.65 T Westport Phyllocolpa sp. Chr13 4.99 0.045 8.82 T Westport Gentisyl alcohol 5-O-glucoside Chr17 12.2 <0.0001 25.760 T Grand Rapids Gentisyl alcohol 5-O-glucoside Chr10 10.3 <0.0001 23.104 T Grand Rapids Gentisyl alcohol 5-O-glucoside Chr14 5.96 0.0100 7.296 T Grand Rapids

Table 3. One-way ANOVA, with Genotype as a covariate, analyzing the effect of infection severity of the genus of one fungus on the infection severity of the competing leaf fungi in 2008.

Factor DF Sum of squares Mean square F-ratio p-value

Dependent Variable- Melampsora sp. leaf rust infection

Genotype 732 1120 1.53 4.69 <0.0001

S. musiva score 3 34.4 11.5 35.2 <0.0001

Residuals 602 197 0.326

Dependent Variable- S. musiva leaf spot infection

Genotype 732 379 0.518 1.77 <0.0001

Melampsora sp. score 3 27.6 9.19 31.5 <0.0001

Residuals 602 176 0.292

Table 4. Number of genes in QTL intervals in parental genomes for biotic associations.

Fungus/Insect QTL Interval # Genes in P. trichocarpa # Genes in P. deltoides

Melampsora sp. Chr04 159 147

S. musiva canker Chr16 71 68

S. musiva leaf spot Chr06 129 132

Phyllocolpa sp. Chr10 146 154

Phyllocolpa sp. Chr13 252 425

M. vagabunda Chr05 129 147

Table 5. Candidate genes under positive selection (Ka/Ks>1) in genetic intervals associated with fungi and insects.

QTL Arabidopsis function P. trichocarpa gene P. deltoides gene Ka/Ks

Chr04 Chitin elicitor receptor Potri.004G005800 Podel.04G004900.1.p 1.35

kinase 1

Cysteine-rich RLK Chr04 Potri.004G012600 Podel.04G010900.1.p 2.23 (RECEPTOR-like protein

kinase)

Chr04 Stigma-specific Stig1 Potri.004G006800 Podel.04G005700.1.p 1.31

family protein

Chr04 Stigma-specific Stig1 Potri.004G007200.1 Podel.04G006100.1.p 1.00

family protein

Chr04 Stigma-specific Stig1 Potri.004G007200 Podel.04G006100.1.p 1.00

family protein

Pentatricopeptide repeat Chr05 Potri.005G038500.1 Podel.05G042200.1.p 1.52 (PPR) superfamily protein

Chr10 Copper amine oxidase Potri.010G088700 Podel.10G084500.1.p 1.19

family protein

Table 6. Tandem duplication profiles for genetic intervals. Number of copies next to species gene name indicates the size of tandem expansion for the gene.

Arabidopsis function P. trichocarpa gene Copy # P. deltoides gene Copy # QTL Chr

ERD (early- responsive to dehydration Potri.004G005900 3 Podel.04G005000.1.p 3 Chr04 stress) family protein

Pentatricopeptide repeat (PPR) superfamily NA 0 Podel.04G010100.1.p 2 Chr04 protein

Peroxidase superfamily protein Potri.004G015300 1 Podel.04G013500.1.p Chr04

Disease resistance protein (TIR-NBS- LRR Potri.005G031900 7 Podel.05G035500.1.p 12 Chr05 class)

Shikimate O- Hydroxycinnamoyl- Potri.005G028000 3 Podel.05G029500.1.p 2 Chr05 transferase

Protein tyrosine kinase (Pkinase_Tyr) Potri.005G030600 4 Podel.05G030600.1.p 2 Chr05 Leucine rich repeat (LRR_8)

Lipoxygenase Potri.005G032400 4 Podel.05G037200.1.p 3 Chr05

Cytochrome P450, family 721, subfamily A, Potri.005G034500 2 Podel.05G039300.1.p 1 Chr05 polypeptide 1

Cytochrome P450, family 76, subfamily G, Potri.005G029200 5 Podel.05G034500.1.p 0 Chr05 polypeptide 1

Cytochrome P450, family 93, subfamily D, Potri.005G037100 2 NA 0 Chr05 polypeptide 1

Cytochrome P450, family 76, subfamily G, NA 0 Podel.05G034500.1.p 4 Chr05 polypeptide 1

Disease resistance protein (TIR-NBS class), Potri.005G032000 1 Podel.05G035800.1.p 12 Chr05 putative

Receptor-like kinase in flowers Potri.005G040200 1 Podel.05G043500.1.p 2 Chr05

Pentatricopeptide repeat (PPR) superfamily Potri.010G083800 3 Podel.10G079300.1.p 3 Chr10 protein

Copper amine oxidase family protein Potri.010G088800 3 NA 0 Chr10

Ankyrin repeat family protein Potri.013G133400 5 Podel.13G142400.1.p 10 Chr13

O-methyltransferase family protein Potri.013G143700 4 Podel.13G181100.1.p 1 Chr13

O-methyltransferase family protein NA 0 Podel.13G144000.1.p 4 Chr13

G-type lectin receptor-like protein kinase Potri.016G102500 4 Podel.16G106900.1.p 4 Chr16

Figure 1. Biotic phenotype symptoms observed on trees including (a) leaf symptoms of Melampsora sp. fungal leaf rust, (b) leaf spot symptoms of the S. musiva fungus, (c) S. musiva canker symptoms,

(d) branch gall created by the M. vagabunda aphid (e) petiole gall created by the P. populitransversus aphid, and (f) Phyllocolpa sp. leaf folding gall.

Figure 2. Collinear genes in P. deltoides (Pd) and P. trichocarpa (Pt) chromosomes, based on

MCScanX alignments. (a) Chr02; (b) Chr05.

Figure 3. Change in number of tandemly duplicated genes discovered with increasing window size.

Figure 4. QTL interval plots showing peaks across the genome that associate with 2008 biotic surveys.

Lines on the plots indicate p-value thresholds as determined by running 1000 permutations of mapping model for (a) Melampsora sp. model with all individuals and clones subset to exclude individuals with

S. musiva infection and (b) S. musiva leaf spot model with all individuals and clones subset to exclude individuals with Melampsora sp. infection.

Figure 5. QTL interval plots showing peaks across the genome that associate with biotic surveys.

Lines on the plots indicate p-value thresholds as determined by running 1000 permutations of mapping model for (a) binary fungal survey of S. musiva canker, (b) binary insect surveys of M. vagabunda and

P. populitransversus, (c) and insect surveys of Phyllocolpa sp. leaf gall counts and overlapping peaks for gentisyl alcohol 5-O-glucoside compound QTL.

Figure 6. Mean and standard error of fungal infection for individuals with varying category levels of competing fungus infection. Letters indicate significantly different means as determined by Tukey's honest significance test for each one-way ANOVA test.

Figure 7. Comparison of gene content in P. trichocarpa grandparent 93-968 (left line) and P. deltoides grandparent Ill-101 (right line) for significant genetic intervals for (a) Melampsora sp. chromosome 4,

(b) S. musiva chromosome 16, (c) M. vagabunda chromosome 5, (d) Phyllocolpa sp. chromosome 10, and (e) Phyllocolpa sp. chromosome 13 associations. QTL intervals were defined as 1 Mb regions centered on the marker with the highest LOD score. Size of gene point is relative to the number of genes in the tandem duplication expansion.

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Figure 7.

Chapter 3: Genetic underpinnings of arthropod community

distributions in Populus trichocarpa

65 3.1 Abstract

Community genetics seeks to understand the mechanisms by which natural genetic variation in plant populations can extend phenotypes to predictably establish entire assemblages of organisms. Until now, much of the research in this field has focused on plant genotypes in naturally occurring hybrid zones of Populus, which show elevated influence on arthropod communities, with less work completed on single species populations and even fewer having identified candidate genes controlling hierarchical ecosystem organization. We surveyed arthropod herbivores and predators on genotypes of an association population of P. trichocarpa planted in three common gardens. Plant genotype was influential in structuring arthropod community composition among all garden sites and years. Additionally, using GWAS and functional networks, we were able to identify several candidate genes mediating plant-arthropod community relationships. Genes associated with occurrence of Phyllonorycter sp. suggest insect manipulation of host tissue nutrition as well as a host resistance response to attack. Candidate genes important for higher-level organization of communities had more broadly applicable functions, such as terpenoid biosynthesis and production of dsRNA binding proteins and protein kinases, that may be capable of targeting multiple arthropod species. Through our research, we have shown that natural genetic variation in a single species, P. trichocarpa, is just as capable of strongly organizing the assemblage of herbivores and predators as hybrid populations.

66 3.2 Introduction

Genetic variation in a plant species is capable of extending beyond traditional phenotypes, ultimately structuring entire assemblages of organisms (Whitham et al., 2003). The field of community genetics links ecology into this evolutionary framework to predict community structure and function as underlying genes shift within plant populations (Whitham et al., 2006;

Hughes et al., 2008). Prior research of these relationships in agricultural systems has shown that plant genotype has a strong effect on herbivore distributions (Barbour et al., 2016; Crutsinger,

2016). However, the impacts of genotype are expected to increase when a species of plant is either dominant or foundational in the environment (Ellison et al., 2005; Whitham et al., 2006).

Given the depth of genetic information on Populus and their dominant role in riparian ecosystems, Populus trees have been a common model to explore the heritability of community composition (Wimp et al., 2005; Shuster et al., 2006; DeWoody et al., 2013).

Arthropod communities of Populus are known to be structured by a number of heritable plant, traits including tissue chemistry, morphology, and phenology (Whitham, 1978; Floate et al.,

1993; Boeckler et al., 2011). Surveys of Populus populations showed that plant genotype is capable of structuring arthropod communities (Wimp et al., 2005), which in turn can cascade out into additional trophic levels such as the distribution of avian predators (Bailey et al., 2006) and fungi (Dickson & Whitham, 1996). Similarly, Populus genotypes influence beaver foraging

(Bailey et al., 2004), which can indirectly benefit leaf feeding beetles (Martinsen et al., 1998). In addition to these multi-trophic interactions, genetic variation in tree populations has also been able to foster cryptic speciation in specialist mites (Evans et al., 2008).

67 A noticeable pattern in early studies is the use of trees from natural Populus hybrid zones for community genetics research. Hybrids tend to display stronger community effects due to higher susceptibility/resistance to fungal and insect associations (Whitham et al., 1999). The hybrid relationship with the tree insect community is so elevated it has been suggested as an alternative mechanism to distinguish closely-related plant taxa in Populus instead of costly genetic markers

(Floate & Whitham, 1995). More recent work has established similar community interactions are also detectable in single species populations of Populus, including structuring arthropod-avian predator distributions (Smith et al., 2011), microbes and lichens (Lamit et al., 2015), as well as aquatic-terrestrial linkages (Crutsinger et al., 2014). However, for many of these studies, the reliance on estimates of genotypic richness, and assumption of equal relatedness among genets, limited the ability to identify the genes underlying arthropod community interactions (Barker et al., 2019; Wimp et al., 2019).

The abundant genetic resources of Populus trichocarpa make it ideal for discovery of genes that contribute to community composition of arthropods. With a full reference genome sequence

(Tuskan et al., 2006) and the sequencing of over 1,100 individual trees collected throughout the species range, a genetic map containing millions of segregating single nucleotide polymorphisms

(SNPs) is available to answer the “genes-to-ecosystem” questions (Evans et al., 2014). Coupling these genetic resources with traditional single-trait genome-wide association analysis (GWAS), new multi-trait GWAS methods (Chhetri et al., 2019), and networks built from a diverse set of plant phenotypes will allow us to discover previously unknown genes contributing to arthropod community composition.

68 In this study, we surveyed arthropod communities across three common gardens and two years to

address three main questions: (1) Is there heritable genetic control of individual arthropods,

calculated community metrics, and arthropod community composition? (2) Which genes underlie

these associations? (3) What possible biological functions could candidate genes confer based on co-expression networks?

3.3 Methods

Association population plantation

A total of 1,100 P. trichocarpa genotypes were previously collected from across the natural range of P. trichocarpa from northern California to northern British Columbia (Slavov et al.,

2012; Evans et al., 2014). These were planted in 2009 in a randomized, three block design with three replicates of each genotype in Clatskanie, OR, Corvallis, OR and Placerville, CA (Figure

1). The trees were planted at a 2 m x 3 m spacing in Corvallis, and 3 m x 3 m in the other two sites.

Arthropod surveys

Surveys were conducted in the Clatskanie, Placerville, and Corvallis plantations in July 2012 and again in the Corvallis plantation in July 2015. The 2012 plantation surveys were performed on five branches of approximately equal size for each tree while surveys conducted in 2015 were performed on ten branches. All insects and leaf modifier damage present on surveyed branches were counted and recorded (Bangert et al., 2008) (Table 1). Surveys completed in

Clatskanie in 2012 consisted of 712 trees and a total of 654 genotypes (36 with replicate observations), in Corvallis 102 trees and 40 genotypes (35 with replicate observations) were

69 observed, and in Placerville 100 trees and 40 genotypes (34 with replicate observations) were

observed. Surveys completed in Corvallis in 2015 consisted of 557 trees and 496 genotypes (61

with replicate observations).

Genotypic data collection

Methods for obtaining genotypic data were described elsewhere previously (Chhetri et al., 2019;

Weighill et al., 2019). Briefly the genotypes were sequenced to a minimum of 15X depth for a

total of 1053 trees using Illumina genetic analyzers at the DOE Joint Genome Institute. We

removed trees that were related more closely than first cousins and removed highly differentiated

California trees. This left a total of 869 trees that were used for the GWAS analyses. We further

removed SNPs with minor allele frequency ≤ 0.05 and markers with severe departures from

Hardy-Weinberg expectations.

3.4 Statistical analysis

Within-garden microsite variation was removed from survey datasets using thin-plate spline

regression (fields R package) for insect species that were abundant in plantations. Residuals of

the spatial correction were added to the mean of each trait to spatially correct values. For

genotypes with replicate observations, the genotypic component of variance in insect traits was

estimated using a linear-mixed model (lmer function of the lme4 R package) with the spatially

corrected insect trait as the response, tree genotype as the predictor, and a stem diameter proxy for biomass (Ruark et al., 1987) as a covariate, where applicable. Broad-sense heritability was

calculated as:

2 2 2 2 H = δg / (δg + δe )

70 2 2 where g is the genetic variance due to genotype and e is the residual variance. Rapid,

simulation𝛿𝛿 -based exact likelihood ratio tests were used𝛿𝛿 to evaluate the significance of variation

due to genotype for each linear model (exactRLRT function of the RLRsim R package). The same

models were also used to estimate heritability of calculated community metrics from the original

dataset including abundance (sum of all insects), richness (total number of unique species), and

Shannon-Weaver diversity.

Arthropod community composition was evaluated using non-metric multidimensional scaling

(NMDS) with Bray-Curtis dissimilarity both across all gardens surveyed for the same 44 genotypes in 2012 and within each garden across sites and years. A PERMANOVA (adonis function of the vegan R package) was run for replicated genotypes to determine the influence of tree genotypes on arthropod community composition among all NMDS configurations.

Coordinates for dependent variables were extracted from the NMDS configurations and depicted in plots using a rescaled font to represent the three-dimensional projection. This was accomplished by rescaling all axis scores to a zero origin and scaling the font size relative to the product of the three rescaled axis scores.

Association and network analysis

Genome-wide association tests were performed using the Genome-wide Efficient Mixed Model

Association package (GEMMA) (Zhou & Stephens, 2012, 2014). The GWAS association test included phenotypic BLUPs, a genetic relationship matrix, and 6,741,160 SNPs. The genetic relationship matrix was also estimated using GEMMA. Single-trait GWAS was performed for insect and community traits from the Clatskanie 2012 surveys (654 genotypes) and Corvallis

71 2015 surveys (496 genotypes) (Table 2). Insect traits including Chrysomela scripta,

Phyllonorycter sp. blotch miner, Phyllocnistis populiella serpentine miner, and Corythucha sp. lace wing counts were binarized for analysis given the non-normal distribution of their occurrence in the field. The tested model was:

y = Wα + xβ + u + ϵ where y is an n-vector of phenotypic BLUP values, where n is the number of individuals tested;

W is an n×c matrix of covariates; α is a c-vector of corresponding coefficients, where c is the number of principal coordinate axes used; x is an n-vector of marker genotypes, β is the effect size of the marker, u is an n-vector of random effects that includes a relatedness matrix and ϵ is an n-vector of errors.

Multitrait GWAS was performed for two multitrait sets, including the NMDS configuration from the Clatskanie 2012 (654 genotypes) and Corvallis 2015 (496 genotypes) sites. We combined 5

NMDS axes obtained from the community genetics analyses independently for each of the plantations in Corvallis and Clatskanie. Multitrait association was conducted with GEMMA using the same model as for single trait associations, except y is an n×d matrix of d phenotypes for n individuals. We used an FDR cutoff of 0.05 and a more liberal FDR cutoff of 0.1 to identify suggestive associations (Storey & Tibshirani, 2003).

To assess possible biological function of candidate genes identified for insect/community metric associations identified from GWAS, networks were constructed for genes from the same surveyed populations. Networks were based on genes that were co-expressed with candidate genes using the JGI Plant Gene Atlas for P. trichocarpa (https://phytozome.jgi.doe.gov),

72 methylation data for multiple P. trichocarpa tissue types (Vining et al., 2012), as well as GWAS for pyMBMS (Weighill et al., 2018) and metabolite profiles as determined by GC-MS

(Tschaplinski et al., 2012). Further detail of network construction can be found in Weighill et al.

(2018). Candidate genes of functional interest to insect traits were selected using strength of SNP association (p-values) and SNP distance from gene determined from GWAS. Networks were merged based on shared nodes. Candidate genes with significant GWAS associations and with depth (number of connected layers) greater than 2 and breadth (number of neighbor genes) greater than 500 in the merged networks were selected for further analysis. Enrichment of GO functional categories in corresponding one-hop subnetworks was tested using Fisher’s Exact

Tests with a significance threshold of FDR ≤ 0.1. Final networks were visualized using

Cytoscape v.3.6.1(Shannon et al., 2003).

3.5 Results

Broad-sense heritability and NMDS configuration

Significant broad-sense heritability was detected for all insect and community traits in one or more of the surveys across gardens and years (Table 2). The abundance of the Harmandia sp. trait was heritable in two garden surveys in 2012 including Clatskanie (H2 = 0.511, p-value

<0.001) and Corvallis (H2 = 0.394, p-value <0.001). In the 2015 Corvallis surveys, only two

insect traits were genetically heritable, including Phyllocolpa sp. (H2 = 0.237, p-value = 0.043)

and Corythucha sp. (H2 = 0.145, p-value = 0.05). Abundance of insects occurring on branches was significantly heritable in Placerville in 2012 (H2 = 0.181, p-value = 0.042) and Corvallis in

2015 (H2 = 0.268, p-value = 0.016). Insect richness was also heritable in Placerville in 2012 (H2

= 0.217, p-value = 0.037). Shannon-Weaver diversity was heritable in Clatskanie in 2012 (H2 =

73 0.231, p-value = 0.05), Placerville in 2012 (H2 = 0.244, p-value = 0.026), and Corvallis in 2012

(H2 = 0.191, p-value = 0.046).

The number of NMDS dimensions required to adequately explain dissimilarities in insect community was 4 across all gardens, and within each garden was 5 for Clatskanie 2012, 3 for

Placerville 2012, 3 for Corvallis in 2012, and 5 for Corvallis in 2015. PERMANOVA revealed significant garden effects across 2012 surveys (Table 3; Figure 2) and significant genetic effects on arthropod community composition among all NMDS configurations (Table 4; Figure 3,4).

GWAS and network results

A total of 71 loci were significantly associated with insect and community traits. P. populiella yielded 21 significant SNPs across 19 loci and 13 chromosomes from the Clatskanie survey

(Figure 5a; Table 5). Phyllonorycter sp. blotch miner yielded a combined total of 31 significant

SNPs across 27 loci and 14 chromosomes; of these, 30 SNPs and 26 loci were discovered from the Clatskanie 2012 survey and 1 SNP and 1 locus was discovered from the Corvallis 2015 survey (Figure 5a; Table 6). Insect richness was associated with a combined total of 11 SNPs across 9 loci on 8 separate chromosomes; of these, 6 SNPs and 5 loci were discovered from the

Clatskanie 2012 survey and 5 SNPs and 4 loci were discovered from the Corvallis 2015 survey.

Insect abundance in the Clatskanie site also yielded 4 significant SNPs across 3 loci and 2 chromosomes (Figure 5b; Table 7). Both the Clatskanie and Corvallis sites yielded significant associations for the insect community composition (Figure 5c; Table 8). There were 11 significant SNPs across 4 loci and 4 chromosomes from the Clatskanie GWAS analysis, while the Corvallis site yielded 13 SNPs across 9 loci and 9 chromosomes. Of the GWAS SNPs which

74 were located on/near 66 P. trichocarpa genes, 6 genes were identified from networks as strong candidates for functional importance (Table 9) including Clatskanie survey of Phyllonorycter sp. blotch miner (Figure 6), Corvallis arthropod richness (Figure 7), and Corvallis community composition (Figure 8). Significant GO term functional enrichments of co-expressed genes from networks were detected for 4 of the 6 candidate genes (Table 10).

3.6 Discussion

Host plant genetics can have predictable effects on community assembly of living organisms at multiple hierarchical levels (Crutsinger et al., 2008, 2009; Barbour et al., 2009), especially in naturally occurring hybrid zones of Populus (Wimp et al., 2005; Whitham et al., 2006, 2012). In our study, we have demonstrated, across surveys of arthropods in multiple gardens and years,

that host genotype has a significant effect on the community of associated arthropods, even with

the unique assemblages of insects at each location. Additionally, we were able leverage the

genomic resources of P. trichocarpa to identify the underlying genes and potential functions that

are important in several levels of arthropod community assembly, including the distribution of

the leaf mining specialist arthropod Phyllonorycter sp., the number of species (richness) capable

of co-existing on a single genotype, and community composition.

Phyllonorycter sp. larvae develop in the leaf mesophyll and their first three early instar stages

rely on sap-feeding while the last two feed on leaf tissue (Davis & Deschka, 2001). Two

candidate genes that were associated with Phyllonorycter sp. in the Clatskanie site appear to both

be important in cell proliferation processes that may impact feeding success of the blotch miner.

A gene located on Chr01 was identified to be cyclin B1;4 and is known to be an important cell

75 cycle regulator in Populus (Dong et al., 2011; Plavcova et al., 2013), with further support for this function in the form of co-expression enrichment in our networks with processes involved in chromosome segregation/organization and multicellular development. Galling and mining insects have been shown to be directly capable of manipulating molecular mechanisms important in cell proliferation via phytohormones (Erb et al., 2012). Furthermore, some species of

Phyllonorycter have already been implicated in indirectly manipulating cytokinin levels in host plant tissues, thereby preventing senescence and increasing the nutritive value of leaves for larval development (Kaiser et al., 2010; Erb et al., 2012).

Interestingly, the same candidate gene on Chr01 was directly associated with two metabolites including the dihydromyricetin. Although dihydromyricetin has been found to be detrimental to the survival of generalist herbivores (Peters & Constabel, 2002), flavonoids are also capable of being important ovipositional stimulants to specialist herbivores (Simmonds, 2001). An additional candidate gene located on Chr18 and identified to be an adenine phosphoribosyl transferase-like protein, may have implications in suppressing cell division. Adenine phosphoribosyl transferases are capable of utilizing cytokinins as substrates, converting them

into inactive intermediate or storage forms (Allen et al., 2002; Plavcova et al., 2013). Moreover, the same gene has been associated with production of alanine, which, when used in the modification of cytokinin, has the ability to permanently deactivate the phytohormone

(Srivastava, 2002), possibly preventing the tissue manipulating behavior of the leaf miners.

Although there were no definitive results for individual insect species with abundant distributions in the Corvallis garden site, we discovered multiple gene associations with two

76 community metrics. Arthropod richness was significantly associated with two potential candidate genes located on Chr07 and Chr18. Arthropod richness refers to the total number of unique species observed to be interacting with plant tissues, including a combination of insect predators as well as specialist and generalist herbivores and damage types. The gene located on Chr07 was identified to be an RNA-binding protein Musashi (MSI). Little is known regarding MSI protein function in plants, but the subnetwork of connected genes was enriched for RNA- and histone- binding functions. RNA-binding proteins are generally important in post-transcriptional gene regulation as they guide RNA stability/transport and affect chromatin modification and also have

implications in abscisic acid hormone signaling (Thaler & Bostock, 2004; Lorković, 2009).

The gene on Chr18 that was associated with insect richness encodes a protein that is similar to a

cytochrome P450 enzyme. Plant cytochrome P450’s are known to have a wide diversity of

biological catalyst functions. In Populus they have been frequently associated with recovery

from abiotic stresses, such as drought and salt (Gu et al., 2004; Chen et al., 2013; Wang et al.,

2016), as well as dealing with biotic stress (Irmisch et al., 2013, 2014a). The same cytochrome

P450 gene was associated with an unidentified metabolite. The associated subnetwork was

enriched for terpenoid (isoprenoid) biosynthetic processes as well as molecular mevalonate

kinase activity. This may provide valuable clues about the possible identity of the metabolite as

well as the function of this uncharacterized gene.

Cytochrome P450 genes are known to be involved in terpenoid biosynthesis in plants (Keeling &

Bohlmann, 2006). The relationship of terpenoids and the number of species observed on plant

tissues is highly dependent on the evolutionary history of the insect with its host. Generalist

77 insects that are not adapted to Populus chemistry will potentially be deterred by

constitutive/induced terpenoid compounds, as consumption of these can have detrimental effects

on their physiology and development (Wada & Munakata, 1971; Huber & Bohlmann, 2004;

Keeling & Bohlmann, 2006). Conversely, specialist insects that have coevolved to utilize

Populus as a host are capable of utilizing terpenoid compounds as oviposition/location cues

(Brilli et al., 2009). Finally, terpenoids released as volatiles from crushed plant tissues have been

found to be important indirect plant defenses as they can aid in the attraction of predators and

parasites of insect herbivores (Huber & Bohlmann, 2004; Irmisch et al., 2014b). This is particularly interesting in the case of our surveys given that most predators present in the garden site were made up of lady bugs (Coccinellidae) and spiders belonging to Salticidae and

Araneidae families. There are many examples of the importance of plant volatiles to lady bug foraging of aphids (Ninkovic et al., 2001; de Vos & Jander, 2010), whereas the relationship between spiders and volatiles has not been as thoroughly explored. However, more literature is being published supporting the importance of volatile compounds in the ability of spiders to locate and hunt prey (Zhao et al., 2002; Nelson et al., 2012).

We also identified two promising candidate genes that were mediating the arthropod community composition. One gene, located on Chr17, is a double-stranded RNA (dsRNA) binding protein. dsRNA-binding proteins have been implicated in RNA interference (RNAi), which is capable of silencing genes in not only herbivore insects but also disease vectors, including fungal pathogens, viruses, and bacteria (Niu et al., 2010; Yoon et al., 2018). The wide diversity of potential targets of dsRNA-binding proteins could enable them to act as a potential first response to biotic stress. Moreover, the subnetwork contained sucrose transporter genes that were

78 negatively co-expressed with the dsRNA binding protein. This could cause a reduction in nutritive value of the tissue, thereby providing a generic response that would reduce attractiveness of the tissue to a broad spectrum of herbivores (Niu et al., 2010).

Another gene associated with community composition was a wall-associated receptor kinase C- terminal gene that acts as a transmembrane protein. This gene family is greatly expanded in

Populus and has been implicated in pathogen responses in several plant systems (Li et al., 2009;

Tocquard et al., 2014; Hurni et al., 2015). Although pathogens were not a part of these surveys, herbivores are capable of predisposing trees to secondary pathogen infection and pathogen infection can weaken plant immune systems making them more susceptible to herbivore feeding

(Busby et al., 2019). As a result, the function of wall-associated receptor kinases may be important as a general response to biotic stress preventing secondary infections due to weakened immune systems and mass attack.

Overall, we have shown across three common gardens, each of which had unique assemblages of arthropod species, that plant genotype within a single species has strong influence over its community of herbivores and predators. We also discovered several candidate genes in P. trichocarpa important in underlying individual herbivore distributions, total number of species on plant tissues, and community composition. Candidates mediating the specialist herbivore

Phyllonorycter sp. blotch miner reflected very targeted responses important in the insect’s ability to manipulate host plant growth and the potential host response to resist attack. Conversely, genes that were associated with species richness and community composition have potential functions, such as terpenoid and dsRNA-binding protein production, that may be capable of

79 more broadly targeting groups of insects. Future work should aim to confirm the functions of these candidates as well as determine how genes of similar function may contribute to community genetics effects in additional pure species of Populus.

80 3.7 Literature cited

Allen M, Qin W, Moreau F, Moffatt B. 2002. Adenine phosphoribosyltransferase isoforms of

Arabidopsis and their potential contributions to adenine and cytokinin metabolism.

Physiologia Plantarum 115: 56–68.

Bailey JK, Schweitzer JA, Rehill BJ, Lindroth RL, Martinsen GD, Whitham TG. 2004.

Beavers as molecular geneticists: a genetic basis to the foraging of an ecosystem

engineer. Ecology 85: 603–608.

Bailey JK, Wooley SC, Lindroth RL, Whitham TG. 2006. Importance of species interactions

to community heritability: a genetic basis to trophic‐level interactions. Ecology Letters 9:

78–85.

Bangert, RK, Lonsdorf, EV, Wimp, GM, Shuster, SM, Fischer, D, Schweitzer, JA,

Whitham, TG. 2008. Genetic structure of a foundation species: scaling community

phenotypes from the individual to the region. Heredity 100: 121-131.

Barbour MA, Fortuna MA, Bascompte J, Nicholson JR, Julkunen-Tiitto R, Jules ES,

Crutsinger GM. 2016. Genetic specificity of a plant–insect food web: implications for

linking genetic variation to network complexity. Proceedings of the National Academy of

Sciences 113: 2128–2133.

Barbour RC, Storer MJ, Potts BM. 2009. Relative importance of tree genetics and

microhabitat on macrofungal biodiversity on coarse woody debris. Oecologia 160: 335–

342.

Barker HL, Riehl JF, Bernhardsson C, Rubert‐Nason KF, Holeski LM, Ingvarsson PK,

Lindroth RL. 2019. Linking plant genes to insect communities: identifying the genetic

bases of plant traits and community composition. Molecular Ecology 28: 4404–4421.

81 Boeckler GA, Gershenzon J, Unsicker SB. 2011. Phenolic glycosides of the Salicaceae and

their role as anti-herbivore defenses. Phytochemistry 72: 1497–1509.

Brilli F, Ciccioli P, Frattoni M, Prestininzi M, Spanedda AF, Loreto F. 2009. Constitutive

and herbivore‐induced monoterpenes emitted by Populus × euroamericana leaves are key

volatiles that orient Chrysomela populi beetles. Plant, Cell & Environment 32: 542–552.

Busby PE, Crutsinger G, Barbour M, Newcombe G. 2019. Contingency rules for pathogen

competition and antagonism in a genetically based, plant defense hierarchy. Ecology and

Evolution 9: 6860–6868.

Chen J, Song Y, Zhang H, Zhang D. 2013. Genome-wide analysis of gene expression in

response to drought stress in Populus simonii. Plant Molecular Biology 31: 946–962.

Chhetri HB, Macaya‐Sanz D, Kainer D, Biswal AK, Evans LM, Chen J, Collins C, Hunt K,

Mohanty SS, Rosenstiel T, et al. 2019. Multitrait genome‐wide association analysis of

Populus trichocarpa identifies key polymorphisms controlling morphological and

physiological traits. New Phytologist 223: 293–309.

Crutsinger GM. 2016. A community genetics perspective: opportunities for the coming decade.

New Phytologist 210: 65–70.

Crutsinger GM, Cadotte MW, Sanders NJ. 2009. Plant genetics shapes inquiline community

structure across spatial scales. Ecology Letters 12: 285–292.

Crutsinger GM, Reynolds WN, Classen AT, Sanders NJ. 2008. Disparate effects of plant

genotypic diversity on foliage and litter arthropod communities. Oecologia 158: 65–75.

Crutsinger GM, Rudman SM, Rodriguez‐Cabal MA, McKown AD, Sato T, MacDonald

AM, Heavyside J, Geraldes A, Hart EM, LeRoy CJ, et al. 2014. Testing a ‘genes‐to‐

ecosystems’ approach to understanding aquatic–terrestrial linkages. Molecular Ecology

82 23: 5888–5903.

Davis DR, Deschka G. 2001. Biology and systematics of the North American Phyllonorycter

leafminers on Salicaceae, with a synoptic catalog of the Palearctic species (Lepidoptera:

Gracillariidae). Smithsonian Contributions to Zoology 614: 1–89.

DeWoody J, Viger M, Lakatos F, Tuba K, Taylor G, Smulders MJ. 2013. Insight into the

genetic components of community genetics: QTL mapping of insect association in a fast-

growing forest tree. PLoS ONE 8: e79925.

Dickson LL, Whitham TG. 1996. Genetically-based plant resistance traits affect arthropods,

fungi, and birds. Oecologia 106: 400–406.

Dong Q, Zhao Y, Jiang H, He H, Zhu S, Cheng B, Xiang Y. 2011. Genome-wide

identification and characterization of the cyclin gene family in Populus trichocarpa.

Plant Cell, Tissue and Organ Culture (PCTOC) 107: 55–67.

Ellison AM, Bank MS, Clinton BD, Colburn EA, Elliott K, Ford CR, Foster DR, Kloeppel

BD, Knoepp JD, Lovett GM, et al. 2005. Loss of foundation species: consequences for

the structure and dynamics of forested ecosystems. Frontiers in Ecology and the

Environment 3: 479–486.

Erb M, Meldau S, Howe GA. 2012. Role of phytohormones in insect-specific plant reactions.

Trends in Plant Science 17: 250–259.

Evans LM, Allan GJ, Shuster SM, Woolbright SA, Whitham TG. 2008. Tree hybridization

and genotypic variation drive cryptic speciation of a specialist mite herbivore. Evolution:

International Journal of Organic Evolution 62: 3027–3040.

Evans LM, Slavov GT, Rodgers-Melnick E, Martin J, Ranjan P, Muchero W, Brunner

AM, Schackwitz W, Gunter L, Chen JG, et al. 2014. Population genomics of Populus

83 trichocarpa identifies signatures of selection and adaptive trait associations. Nature

Genetics 46: 1089.

Floate KD, Kearsley MJ, Whitham TG. 1993. Elevated herbivory in plant hybrid zones:

Chrysomela confluens, Populus and phenological sinks. Ecology 74: 2056–2065.

Floate KD, Whitham TG. 1995. Insects as traits in plant systematics: their use in discriminating

between hybrid cottonwoods. Canadian Journal of Botany 73: 1–13.

Gu R, Fonseca S, Puskás LG, Hackler Jr L, Zvara Á, Dudits D, Pais MS. 2004. Transcript

identification and profiling during salt stress and recovery of Populus euphratica. Tree

Physiology 24: 265–276.

Huber DP, Bohlmann J. 2004. Terpene synthases and the mediation of plant-insect ecological

interactions by terpenoids: a mini-review. In: Cronk Q, Whitton J, Ree R, Taylor I, eds.

Plant adaptation: molecular genetics and ecology. Vancouver, Canada: National Research

Council Canada, 70–81.

Hughes AR, Inouye BD, Johnson MT, Underwood N, Vellend M. 2008. Ecological

consequences of genetic diversity. Ecology Letters 11: 609–623.

Hurni S, Scheuermann D, Krattinger SG, Kessel B, Wicker T, Herren G, Fitze MN, Breen

J, Presterl T, Ouzunova M, et al. 2015. The maize disease resistance gene Htn1 against

northern corn leaf blight encodes a wall-associated receptor-like kinase. Proceedings of

the National Academy of Sciences 112: 8780–8785.

Irmisch S, Clavijo McCormick A, Günther, J., Schmidt A, Boeckler GA, Gershenzon J,

Unsicker SB, Köllner TG. 2014a. Herbivore‐induced poplar cytochrome P450 enzymes

of the CYP 71 family convert aldoximes to nitriles which repel a generalist caterpillar.

The Plant Journal 80: 1095–1107.

84 Irmisch S, Jiang Y, Chen F, Gershenzon J, Köllner T. 2014b. Terpene synthases and their

contribution to herbivore-induced volatile emission in western balsam poplar (Populus

trichocarpa). BMC Plant Biology 14: 270.

Irmisch S, McCormick AC, Boeckler GA, Schmidt A, Reichelt M, Schneider B, Block K,

Schnitzler JP, Gershenzon J, Unsicker SB, et al. 2013. Two herbivore-induced

cytochrome P450 enzymes CYP79D6 and CYP79D7 catalyze the formation of volatile

aldoximes involved in poplar defense. The Plant Cell 25: 4737–4754.

Kaiser W, Huguet E, Casas J, Commin C, Giron D. 2010. Plant green-island phenotype

induced by leaf-miners is mediated by bacterial symbionts. Proceedings of the Royal

Society B: Biological Sciences 277: 2311–2319.

Keeling CI, Bohlmann J. 2006. Genes, enzymes and chemicals of terpenoid diversity in the

constitutive and induced defence of conifers against insects and pathogens. New

Phytologist 170: 657–675.

Lamit LJ, Busby PE, Lau MK, Compson ZG, Wojtowicz T, Keith AR, Zinkgraf MS,

Schweitzer JA, Shuster SM, Gehring CA, et al. 2015. Tree genotype mediates

covariance among communities from microbes to lichens and arthropods. Journal of

Ecology 103: 840–850.

Li H, Zhou SY, Zhao WS, Su SC, Peng YL. 2009. A novel wall-associated receptor-like

protein kinase gene, OsWAK1, plays important roles in rice blast disease resistance. Plant

Molecular Biology 69: 337–346.

Lorković ZJ. 2009. Role of plant RNA-binding proteins in development, stress response and

genome organization. Trends in Plant Science 14: 229–236.

Martinsen GD, Driebe EM, Whitham TG. 1998. Indirect interactions mediated by changing

85 plant chemistry: beaver browsing benefits beetles. Ecology 79: 192–200.

Nelson XJ, Pratt AJ, Cheseto X, Torto B, Jackson RR. 2012. Mediation of a plant-spider

association by specific volatile compounds. Journal of Chemical Ecology 38: 1081–1092.

Ninkovic V, Al Abassi S, Pettersson J. 2001. The influence of aphid-induced plant volatiles on

ladybird beetle searching behavior. Biological Control 21: 191–195.

Niu JH, Jian H, Xu JM, Guo YD, Liu Q. 2010. RNAi technology extends its reach:

engineering plant resistance against harmful eukaryotes. African Journal of

Biotechnology 9: 7573–7582.

Peters DJ, Constabel CP. 2002. Molecular analysis of herbivore‐induced condensed tannin

synthesis: cloning and expression of dihydroflavonol reductase from trembling aspen

(Populus tremuloides). The Plant Journal 32: 701–712.

Plavcova L, Hacke UG, Almeida‐Rodriguez AM, Li E, Douglas CJ. 2013. Gene expression

patterns underlying changes in xylem structure and function in response to increased

nitrogen availability in hybrid poplar. Plant, Cell & Environment 36: 186–199.

Ruark, GA, Martin, GL, Bockheim, JG. 1987. Comparison of constant and variable allometric

ratios for estimating Populus tremuloides biomass. Forest Science 33: 294-300.

Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, Amin N, Schwikowski B,

Ideker T. 2003. Cytoscape: a software environment for integrated models of

biomolecular interaction networks. Genome Research 13: 2498–2504.

Shuster SM, Lonsdorf EV, Wimp GM, Bailey JK, Whitham TG. 2006. Community

heritability measures the evolutionary consequences of indirect genetic effects on

community structure. Evolution 60: 991–1003.

Simmonds MS. 2001. Importance of flavonoids in insect–plant interactions: feeding and

86 oviposition. Phytochemistry 56: 245–252.

Slavov G, DiFazio S, Martin J, Schackwitz W, Muchero W, Rodgers-Melnick E, Lipphardt

M, Pennacchio C, Hellsten U, Pennacchio L, et al. 2012. Genome resequencing reveals

multiscale geographic structure and extensive linkage disequilibrium in the forest tree

Populus trichocarpa. New Phytologist 196: 713–725.

Smith DS, Bailey JK, Shuster SM, Whitham TG. 2011. A geographic mosaic of trophic

interactions and selection: trees, aphids and birds. Evolutionary Biology 24: 422–429.

Srivastava LM. 2002. Cytokinins. In: Plant growth and development: hormones and

environment. San Diego: Academic Press, 191–202.

Storey JD, Tibshirani R. 2003. Statistical significance for genomewide studies. Proceedings of

the National Academy of Sciences 100: 9440–9445.

Thaler JS, Bostock RM. 2004. Interactions between abscisic‐acid‐mediated responses and plant

resistance to pathogens and insects. Ecology 85: 48–58.

Tocquard K, Lafon-Placette C, Auguin D, Muries B, Bronner G, Lopez D, Fumanal B,

Franchel J, Bourgerie S, Maury S, et al. 2014. In silico study of wall-associated kinase

family reveals large-scale genomic expansion potentially connected with functional

diversification in Populus. Tree Genetics & Genomes 10: 1135–1147.

Tschaplinski TJ, Standaert RF, Engle NL, Martin MZ, Sangha AK, Parks JM, Smith JC,

Samuel R, Jiang N, Pu Y, et al. 2012. Down-regulation of the caffeic acid O-

methyltransferase gene in switchgrass reveals a novel monolignol analog. Biotechnology

for Biofuels 5: 71.

Tuskan GA, DiFazio S, Jansson S, Bohlmann J, Grigoriev I, Hellsten U, Putnam N, Ralph

S, Rombauts S, Salamov A, et al. 2006. The genome of black cottonwood, Populus

87 trichocarpa (Torr. & Gray). Science 313: 1596–1604.

Vining KJ, Pomraning KR, Wilhelm LJ, Priest HD, Pellegrini M, Mockler TC, Freitag M,

Strauss SH. 2012. Dynamic DNA cytosine methylation in the Populus trichocarpa

genome: tissue-level variation and relationship to gene expression. BMC Genomics 13:

27.

de Vos M, Jander G. 2010. Volatile communication in plant-aphid interactions. Current

Opinion in Plant Biology 13: 366–371.

Wada K, Munakata K. 1971. Feeding inhibitory activity of terpenoids in plants. Agricultural

and Biological Chemistry 35: 115–118.

Wang C, Yang Y, Wang H, Ran X, Li B, Zhang J, Zhang H. 2016. Ectopic expression of a

cytochrome P450 monooxygenase gene PtCYP714A3 from Populus trichocarpa reduces

shoot growth and improves tolerance to salt stress in transgenic rice. Plant Biotechnology

Journal 14: 1838–1851.

Weighill D, Jones P, Shah M, Ranjan P, Muchero W, Schmutz J, Sreedasyam A, Macaya-

Sanz D, Sykes R, Zhao N, et al. 2018. Pleiotropic and epistatic network-based

discovery: integrated networks for target gene discovery. Frontiers in Energy Research 6:

30.

Weighill D, Macaya-Sanz D, DiFazio SP, Joubert W, Shah M, Schmutz J, Sreedasyam A,

Tuskan G, Jacobson D. 2019. Wavelet-based genomic signal processing for centromere

identification and hypothesis generation. Frontiers in Genetics 10: 487.

Whitham TG. 1978. Habitat selection by Pemphigus aphids in response to response limitation

and competition. Ecology 59: 1164–1176.

Whitham TG, Bailey JK, Schweitzer JA, Shuster SM, Bangert RK, LeRoy CJ, Lonsdorf

88 EV, Allan GJ, DiFazio SP, Potts BM, et al. 2006. A framework for community and

ecosystem genetics: from genes to ecosystems. Nature Reviews Genetics 7: 510–523.

Whitham TG, Gehring CA, Lamit LJ, Wojtowicz T, Evans LM, Keith AR, Smith DS. 2012.

Community specificity: life and afterlife effects of genes. Trends in Plant Science 17:

271–281.

Whitham TG, Martinsen GD, Keim P, Floate KD, Dungey HS, Potts BM. 1999. Plant hybrid

zones affect biodiversity: tools for a genetic‐based understanding of community structure.

Ecology 80: 416–428.

Whitham TG, Young WP, Martinsen GD, Gehring CA, Schweitzer JA, Shuster SM, Wimp

GM, Fischer DG, Bailey JK, Lindroth RL, et al. 2003. Community and ecosystem

genetics: a consequence of the extended phenotype. Ecology 84: 559–573.

Wimp, GM, Tomasula, J, Hamilton, MB. 2019. Putting the genes into community genetics.

Molecular Ecology 28: 4351-4353.

Wimp GM, Martinsen GD, Floate KD, Bangert RK, Whitham TG. 2005. Plant genetic

determinants of arthropod community structure and diversity. Evolution 59: 61–69.

Yoon JS, Mogilicherla K, Gurusamy D, Chen X, Chereddy SC, Palli SR. 2018. Double-

stranded RNA binding protein, Staufen, is required for the initiation of RNAi in

coleopteran insects. Proceedings of the National Academy of Sciences 115: 8834–8339.

Zhao D, Zong-mao C, Cheng J. 2002. Isolation and activity identification of volatiles among

tea plant-green leafhopper-Evarcha spider. Journal of Tea Science 22: 109–114.

Zhou X, Stephens M. 2012. Genome-wide efficient mixed-model analysis for association

studies. Nature Genetics 44: 821–824.

Zhou X, Stephens M. 2014. Efficient algorithms for multivariate linear mixed models in

89 genome-wide association studies. Nature Methods 11: 407–409.

90 3.8 Tables and Figures

Table 1. Sum of arthropods observed on tree tissue within plantations.

Order Arthropod trait Clatskanie Corvallis Placerville Corvallis 2012 2012 2012 2015 Araneae 0 0 0 59 Araneae Salticidae 0 0 0 13 Chrysomela scripta 642 0 0 0 Coccinellidae 0 0 0 55 Coleoptera Polydrusus impressifrons 0 0 0 34 Curculionidae 0 0 0 18 Agromyzid fly 5 0 0 0 Diptera Harmandia sp. 5272 1492 0 0 Dasineura sp. 0 0 0 167 Corythucha sp. 0 0 0 417 Cicadellidae 0 0 0 51 Hemiptera Chloretic damage 0 0 0 108 Mid-vein leaf damage 0 0 0 192 Phyllocolpa sp. leaf fold 1043 297 17 6138 Hymenoptera Phyllocolpa sp. larvae 0 0 0 14 Sawfly larvae 0 0 0 8 Chewing damage 943 143 512 0 Egg mass 628 0 0 16 Leaf damage Holes in leaves 1333 118 538 0 Bottom of leaf damage 1802 21 68 183 Top of leaf damage 259 22 6 0 Leaf folder 3 0 0 0 Leaf modifier Leaf roller 2 0 0 0 Tip folder 63 6 5 0 Phyllonorycter sp. 96 216 0 496 Gypsonoma sp. 38 0 0 10 Lepidoptera Leaf miner 191 0 0 0 Phyllocnistis populiella 310 8 37 7 Sarrothripus frigidana 0 0 0 19

91 Table 2. Linear-mixed model output for community surveys across all gardens and years.

Broad-sense heritability (H2) denotes the contribution of all host plant genetic factors to total variance in phenotype.

# genotypes Garden # genotypes with Phenotype H2 Gvar Evar p-value replicates Clatskanie 2012 654 36 Diversity 0.231 0.034 0.113 0.050

Harmandia sp. 0.511 37.6 36.1 <0.001

Placerville 2012 40 34 Abundance 0.181 10.4 47.1 0.042

Richness 0.217 0.126 0.453 0.037

Diversity 0.244 0.024 0.075 0.026

Corvallis 2012 40 35 Abundance 0.363 95.2 166 <0.001

Diversity 0.191 0.033 0.141 0.046

Harmandia sp. 0.394 84.0 129.3 <0.001

Corvallis 2015 496 61 Abundance 0.268 28.8 78.4 0.016

Corythucha sp. 0.140 0.493 2.90 0.050

Phyllocolpa sp. 0.237 13.9 44.7 0.043

92 Table 3. PERMANOVA results for NMDS configuration within all 2012 gardens with replicated genotype observations and block to account for spatial position.

Factor Df Sum of Mean F-model R2 p-value squares square Garden 2 24.6 12.3 65.1 0.287 0.001

Block 7 3.39 0.484 2.56 0.039 0.001

Genotype 44 13.5 0.307 1.62 0.157 0.001

Residuals 235 44.4 0.189 -- 0.517 --

93 Table 4. PERMANOVA results for NMDS configuration among gardens with replicated genotype observations and block to account for spatial position.

Garden Factor Df Sum of Mean F-model R2 p-value squares squares Block 2 0.790 0.395 1.70 0.035 0.044 Clatskanie 2012 Genotype 34 9.77 0.287 1.23 0.435 0.029

Residuals 51 11.9 0.233 -- 0.529 --

Corvallis Block 2 0.360 0.180 1.14 0.016 0.298 2012 Genotype 39 12.0 0.307 1.95 0.549 0.001

Residuals 60 9.48 0.158 -- 0.434 --

Placerville Block 3 2.24 0.745 5.43 0.132 0.001 2012 Genotype 39 7.07 0.181 1.32 0.416 0.036

Residuals 56 7.68 0.137 -- 0.452 --

Corvallis Block 1 0.185 0.185 1.37 0.010 0.189 2015 Genotype 60 10.1 0.168 1.24 0.549 0.032

Residuals 60 8.11 0.135 -- 0.441 --

94 Table 5. Significant SNPs (FDR < 0.1) and nearest P. trichocarpa genes discovered through GWAS for P. populiella trait in

Clatskanie garden.

Gene Chr Position p-value Description Potri.001G166800 Chr01 14,020,499 3.29*10-7 Von Willebrand factor, Type A domain containing Potri.001G166900 Chr01 14,023,753 7.05*10-8 Similar to hypothetical protein Potri.001G385600 Chr01 40,092,552 4.23*10-8 Protein tyrosine kinase (Pkinase_Tyr) // Di-glucose binding within endoplasmic reticulum (Malectin) Potri.002G244000 Chr02 23,611,478 1.56*10-7 Protein XRI1 Potri.003G062600 Chr03 9,092,517 6.24*10-7 O-fucosyltransferase family protein Potri.003G062700 Chr03 9,096,040 1.15*10-7 Similar to pentatricopeptide (PPR) repeat-containing protein Potri.004G007700 Chr04 497,265 1.22*10-7 Mitogen-activated protein kinase kinase kinase 17-related Potri.005G033500 Chr05 2,494,360 3.95*10-7 Glycogenin glucosyltransferase Potri.005G039300 Chr05 2,864,214 2.85*10-7 Tropinone reductase II / Tropinone reductase Potri.005G170000 Chr05 18,393,885 4.94*10-7 Kelch repeat-containing F-box family protein Potri.005G184700 Chr05 20,171,994 8.47*10-7 Protein stay-green like, Chloroplastic Potri.006G187400 Chr06 20,199,567 6.10*10-7 Polyadenylation factor I complex, subunit FIP1 Potri.006G204100 Chr06 21,938,274 3.40*10-7 No domains Potri.006G235700 Chr06 24,505,808 6.06*10-7 Nuclear poly(A) polymerase Potri.008G020900 Chr08 1,094,017 4.16*10-7 Similar to expressed protein in Arabidopsis thaliana Potri.011G027100 Chr11 2,218,904 4.24*10-7 NADPH-dependent diflavin oxidoreductase 1 Potri.012G046000 Chr12 4,224,452 3.00*10-7 Similar to expressed protein in Arabidopsis thaliana Potri.014G118900 Chr14 9,233,521 7.22*10-7 No domains Potri.015G103600 Chr15 12,174,850 4.16*10-7 Nitrate, fromate, iron dehydrogenase Potri.019G020000 Chr19 2,300,230 4.13*10-7 Similar to Putative peroxisome assembly protein 12 (Peroxin-12) Potri.T148400 scaffold_495 17,502 5.37*10-7 No domains

95 Table 6. Significant SNPs (FDR < 0.1) and nearest P. trichocarpa genes discovered through GWAS for Phyllonorycter sp. trait.

Trait Gene Chr Position p-value Description Phyllonorycter sp. Potri.001G103400 Chr01 8,238,390 3.37*10-7 Phosphatidylinositol N-acetylglucosaminlytransferase subunit P-like protein (Clatskanie) Potri.001G179200 Chr01 15,465,196 3.16*10-7 Coatomer, subunit beta Potri.001G179300 Chr01 15,491,815 2.37*10-8 DNA-directed RNA polymerase V subunit 5C Potri.001G190500 Chr01 1,715,073 1.08*10-7 Calmodulin-binding family protein Potri.001G190600 Chr01 17,160,294 4.93*10-8 No domains Potri.001G190700 Chr01 17,171,087 3.23*10-7 Similar to expressed protein in Arabidopsis thaliana Potri.001G272000 Chr01 27,936,516 4.24*10-7 Cyclin-B1-4 Potri.001G470700 Chr01 50,239,765 7.62*10-7 Similar to Ferredoxin 2; similar to chloroplast precursor. Potri.002G202600 Chr02 16,493,267 6.11*10-7 Pectinesterase family protein Potri.003G116800 Chr03 14,025,351 6.37*10-7 Calcium-binding protein CML24-related Potri.003G204900 Chr03 20,594,059 2.99*10-7 Outer envelope protein 80, Chloroplastic Potri.005G088400 Chr05 6,614,910 4.01*10-7 Gem-like protein 4-related Potri.005G088500 Chr05 6,622,054 1.02*10-7 No domains Potri.005G098600 Chr05 7,507,025 8.55*10-7 Leucine-rich repeat family protein Potri.006G053400 Chr06 3,867,149 3.36*10-7 Similar to guanylate kinase 1 Potri.006G053500 Chr06 3,888,118 2.72*10-7 Transcription factor NF-Y alpha-related Potri.006G053600 Chr06 3,897,089 4.23*10-7 No domains Potri.006G147800 Chr06 12,715,987 3.87*10-7 Similar to casein kinase Potri.006G223400 Chr06 23,563,498 5.91*10-7 No domains Potri.007G093200 Chr07 11,997,824 9.12*10-7 No domains Potri.011G158300 Chr11 17,552,928 1.09*10-7 FAD-binding berberine family protein-related Potri.012G126300 Chr12 14,449,335 8.04*10-7 No domains Potri.013G108900 Chr13 12,228,211 9.49*10-7 ZF-HD protein dimerisation region (ZF-HD_dimer) Potri.013G126100 Chr13 13,887,304 9.52*10-7 Cysteine protease family C1-related Potri.014G130600 Chr14 9,962,648 2.33*10-7 Transducin family protein Potri.014G182800 Chr14 15,527,094 5.27*10-8 Glucan endo-1,3-beta-glucosidase 7-related Potri.015G120200 Chr15 13,506,582 1.90*10-7 Similar to lectin Potri.016G134100 Chr16 13,748,464 5.44*10-8 Co-expressed with genes in roots specific co-expression subnetwork Potri.018G026200 Chr18 2,090,844 2.10*10-8 Similar to Adenine phosphoribosyltransferase 2 Potri.018G076400 Chr18 10,190,180 8.34*10-7 Copper-binding family protein; similar to copper homeostasis factor Phyllonorycter sp. Potri.019G076500 Chr19 11,061,377 2.61*10-8 Domain of unknown function (DUF313) (DUF313) (Corvallis)

96 Table 7. Significant SNPs (FDR < 0.1) and closest P. trichocarpa genes discovered through GWAS for community metric traits.

Trait Gene Chr Position p-value Description Richness Potri.001G068000 Chr01 5,421,300 1.32*10-7 Retinaldehyde binding protein-related (Clatskanie) Potri.001G115500 Chr01 9,243,933 7.23*10-8 ATP-dependent CLP protease Potri.006G038400 Chr06 2,690,405 1.17*10-7 similar to ferredoxin-dependent glutamate synthase Potri.008G079700 Chr08 4,968,425 1.59*10-7 Conserved oligomeric golgi complex subunit 1 Potri.008G079900 Chr08 4,989,054 2.19*10-7 Do-like 15 protein-related Potri.014G116100 Chr14 9,044,620 3.12*10-7 Zinc finger FYVE domain containing protein Richness Potri.007G044400 Chr07 3,884,609 7.77*10-8 RNA-binding protein Musashi (MSI) (Corvallis) Potri.007G044500 Chr07 3,890,366 2.00*10-8 CGI-141-related/lipase containing protein Potri.010G095000 Chr10 11,746,693 3.34*10-7 No domains Potri.013G119400 Chr13 13,255,200 8.10*10-7 TCP family transcription factor (TCP) Potri.018G149300 Chr18 16,746,330 7.48*10-8 Similar to cytochrome P450 family

Abundance Potri.001G243600 Chr01 25,479,678 2.92*10-7 DNA-binding domain-containing protein-related (Clatskanie) Potri.001G243700 Chr01 25,480,842 3.40*10-8 No domains Potri.009G027700 Chr09 3,866,118 2.83*10-7 Similar to responsive to high light 41 Potri.009G028800 Chr09 3,949,481 1.48*10-7 Similar to osmotin-like protein

97 Table 8. Significant SNPs (FDR < 0.1) and closest P. trichocarpa genes discovered through GWAS for arthropod community composition multitrait.

Trait Gene Chr Position p-value Description Community Potri.003G028500 Chr03 3,553,415 2.88*10-7 Adhesion regulating molecule 1 110 KDA cell membrane composition Potri.003G028500 Chr03 3,555,411 5.56*10-7 glycoprotein (Clatskanie) Potri.003G028500 Chr03 3,555,593 4.27*10-7 Potri.003G028500 Chr03 3,555,643 1.96*10-7 Potri.003G028500 Chr03 3,555,663 7.72*10-7 Potri.003G028500 Chr03 3,555,755 5.90*10-7 Potri.003G028500 Chr03 3,555,770 1.70*10-7 Potri.003G028500 Chr03 3,556,231 8.56*10-7 Potri.014G080100 Chr14 6,406,993 1.24*10-7 K15095 - (+)-neomenthol dehydrogenase Potri.015G029200 Chr15 2,364,978 4.13*10-7 Nucleotide-diphospho-sugar transferase domain-containing protein- related Potri.018G131600 Chr18 15,288,689 7.27*10-7 Peroxidase Community Potri.001G199700 Chr01 19,388,433 5.10*10-8 Similar to histidyl-tRNA synthetase composition Potri.001G199700 Chr01 19,388,444 1.34*10-7 (Corvallis) Potri.001G199700 Chr01 19,388,477 5.92*10-8 Potri.002G202800 Chr02 16,515,234 9.06*10-7 Similar to 60S ribosomal protein L18 Potri.008G067900 Chr08 4,132,807 4.48*10-9 ABC1 family protein Potri.010G040500 Chr10 6,904,193 1.20*10-7 Wall-associated receptor kinase galacturonan-binding (GUB_WAK_bind) Potri.010G103900 Chr10 12,466,417 9.33*10-7 Similar to expressed protein in Arabidopsis thaliana Potri.012G005900 Chr12 480,690 4.17*10-7 Two-component response regulator-like APRR5-related Potri.017G063700 Chr17 6,029,280 1.76*10-8 Eukaryote specific dsRNA binding protein Potri.017G063700 Chr17 6,033,420 2.15*10-7 Potri.017G063700 Chr17 6,036,353 7.81*10-7 Potri.017G124700 Chr17 13,673,803 4.29*10-7 Similar to GGPP synthase 2 Potri.T112800 scaffold_178 21,170 2.10*10-9 Protein kinase domain (Pkinase)

98 Table 9. Candidate genes for mediating biotic interactions selected for functional analysis from networks. Breadth indicates number of connected network layers, depth refers to the number of neighbor genes in network connected to anchor gene, and edges indicate the number of connections (co-methylation, co-expression, metabolite GWAS and pyMBMS GWAS) among genes in networks.

Trait Gene Chr Pos Breadth Depth Edges

Clatskanie Potri.001G272000 Chr01 27,936,516 4 1196 1750 Phyllonorycter sp. Clatskanie Potri.018G026200 Chr18 2,090,844 4 615 757 Phyllonorycter sp. Corvallis richness Potri.007G044400 Chr07 3,884,609 3 662 2133 Corvallis richness Potri.018G149300 Chr18 16,746,330 4 1610 6074

Corvallis community Potri.017G063700 Chr17 6,029,280 3 1725 5758

Corvallis community Potri.T112800 scaffold_178 21,170 3 710 1418

99 Table 10. Genome function enrichment analysis of co-expressed neighbor genes from network analysis. Significance of enrichment was determined based on Fisher Exact p-value with multiple testing correction to meet a threshold of FDR < 0.1.

FDR Gene Phenotype Expression GO namespace Function p-value Potri.001G272000 Phyllonorycter Coex-pos Biological process 6.38*10-2 Meiotic chromosome segregation sp. (Clatskanie) 3.97*10-21 Microtubule-based movement 9.29*10-5 Protein polymerization 5.22*10-3 Malate metabolic process 3.86*10-4 Microtubule-based process 2.40*10-2 Proteolysis involved in cellular protein catabolic process Molecular function 1.91*10-2 L-malate dehydrogenase activity 1.98*10-21 Microtubule motor activity 2.22*10-2 Threonine-type endopeptidase activity 9.95*10-5 GTPase activity 1.34*10-3 GTP binding 2.65*10-2 ATP binding Potri.007G044400 Richness Coex-pos Biological process 1.35*10-2 Histone lysine methylation (Corvallis) Molecular function 8.58*10-2 mRNA guanylyltransferase activity 1.36*10-1 Polynucleotide adenylyltransferase activity 1.69*10-1 Histone-lysine N- methyltransferase activity 1.04*10-1 Histone binding 1.26*10-6 Helicase activity 6.92*10-4 Nucleic acid binding 1.43*10-1 RNA binding 0.0905 Zinc ion binding 1.30*10-5 Protein binding 1.42*10-1 DNA binding Potri.018G149300 Richness Coex-pos Biological process 3.87*10-2 Terpenoid (Isoprenoid) (Corvallis) biosynthetic process 2.80*10-2 Phosphorylation Molecular function 2.82*10-3 Mevalonate kinase activity Potri.T112800 Community Coex-pos Biological process 1.80*10-2 Oxidation-reduction process composition Molecular function 6.21*10-2 Ferredoxin-NAD(P) reductase (Corvallis) activity 7.78*10-2 Sigma factor activity 9.63*10-2 Antioxidant activity 3.04*10-2 Oxidoreductase activity Potri.017G063700 Community Coex-neg Biological process 1.55*10-2 Sucrose metabolic process composition (Corvallis)

100 Figure 1. Map displaying genotype origin populations and survey sites – circles represent

collection locations in the wild, stars represent common gardens, the survey sites.

Figure 2. Arthropod community NMDS plot depicting 3 dimensions within all gardens surveyed

in 2012. Color groupings indicate tree surveyed in specified garden.

Figure 3. Arthropod community NMDS plots depicting 3 dimensions for (a) 2012 Clatskanie

garden survey and (b) 2012 Placerville garden survey. PERMANOVA analysis, run with only

genotypes with replicate observations, output indicated for each garden in the upper right corner

of each plot for each common garden. Size of species text indicates position relative to the

dimensional space, for example, larger font indicates forward projection along the positive

values of axes.

Figure 4. Arthropod community NMDS plots depicting 3 dimensions for (a) 2012 Corvallis garden survey and (b) 2015 Corvallis garden survey. PERMANOVA analysis, run with only genotypes with replicate observations, output indicated for each garden in the upper right corner of each plot for each common garden. Size of species text indicates position relative to the dimensional space, for example, larger font indicates forward projection along the positive values of axes.

Figure 5. Manhattan plot output from GWAS for (a) individual arthropods (single-trait; three tests), (b) calculated community metrics for each garden (single-trait; three tests), and (c) community composition (multi-trait; two tests). SNPs that passed the red dotted line were

101 significantly associated with indicated trait. Clatskanie community (c) did not pass threshold but

contained multiple suggestive markers.

Figure 6. Co-expression network for Clatskanie Phyllonorycter sp. single-trait gene candidates.

Co-expression genes were grouped based on their biological function as determined by Gene-

ontology terms.

Figure 7. Co-expression network for Corvallis arthropod richness single-trait gene candidates.

Co-expression genes were grouped based on their biological function as determined by Gene-

ontology terms.

Figure 8. Co-expression network for Corvallis community composition multi-trait gene candidates. Co-expression genes were grouped based on their biological function as determined by Gene-ontology terms.

102 Figure 1.

103 Figure 2.

104 Figure 3.

105 Figure 4.

106 Figure 5.

107 Figure 6.

108 Figure 7.

109 Figure 8.

110 Chapter 4: Characterization of Salix nigra floral insect community

and activity of three native Andrena bees

Submitted to bioRxiv: Simon S, Keefover-Ring K, Park YL, Wimp G, Grady J, & DiFazio S. 2020. Characterization of Salix nigra floral insect community and activity of three native Andrena bees. Authors’ contributions: S.S., S.D., P.Y.-L., and J.G. designed and supervised insect surveys/identifications. K.K.-R and J.G. collected/analyzed metabolite and VOC data. S.S., G.M., and S.D. analyzed the data. S.S., S.D., K.K.-R, P.Y.-L., and G.M. wrote the manuscript.

111

4.1 Abstract

Salix nigra (black willow) is a widespread tree that hosts many species of polylectic hymenopterans and oligolectic bees of the genus Andrena. The early flowering time of S. nigra makes it an important nutritive resource for arthropods emerging from hibernation. However, since S. nigra is dioecious, not all insect visits will lead to successful pollination. Using both visual observation and pan-trapping, we characterized the community of arthropods that visited

S. nigra flowers and assessed the differences among male and female trees as well as the chemical and visual drivers that influenced community composition across three years. We found that male trees consistently supported higher diversity of insects than female trees and only three insect species, all Andrena spp., consistently visited both sexes. Additionally, visits by

A. nigrae correlated strongly to volatile cues suggesting that cross-pollinators cue into specific aspects of floral scent, but diversity of floral visitors is driven strongly by visual cues. Through time, the floral activity of two Andrena species remained stable, but A. nigrae visited less in

2017 when flowers bloomed earlier than other years. When native bee emergence does not synchronize with bloom time, activity appears to be greatly diminished, which could threaten species that only subsist on a single host. Despite the community diversity of S. nigra flowers, productivity depends on a small fraction of insect species that are not threatened by competition, but rather rapidly changing climate conditions that lead to host-insect asynchrony.

112

4.2 Introduction

Early spring emergence of flowers is extremely important in supplying nutritive rewards, such as

pollen and nectar, for many native arthropods, while the host plant benefits with an increased

chance of successful sexual reproduction. The most common cross-pollinators in agricultural systems are often made up of arthropods that collect pollen from many unrelated host plants, such as flies belonging to the family Syrphidae, eusocial bee species such as honey bees (Apis mellifera L.), and polylectic solitary bees (Ostaff et al., 2015). Lack of discrimination among hosts allows these arthropod groups to more flexibly collect resources for survival and population growth.

Conversely, oligolectic solitary bees, which only collect pollen from either related plant species or a single species, rely heavily on predictable timing of available native floral resources

(Danforth, 2007; Straka et al., 2014). Upon emergence from nests, oligolectic bees must locate flowers, feed, breed, build new nests, lay eggs, and collect resources to provide their larvae with food provisioned for development throughout the remainder of the year, all within the bloom time of their specific host (Stevens, 1949; Linsley, 1958; Danforth, 2007). In addition to being a valuable resource for early emerging generalist floral insects, willow species belonging to the genus Salix are the primary hosts of many oligolectic bees, especially those belonging to one of the largest bee genera, Andrena (Stevens, 1949; Ostaff et al., 2015).

Salix encompasses between 300-400 species of shrubs and trees with a dual pollination system, which occurs via wind (anemophilous), insects (entomophilous), or both (Tamura & Kudo, 2000;

Karrenberg et al., 2002; Argus, 2011). Salix biology creates a unique environment for insect

113 reward collection due to its dioecious nature. In order for sexual reproduction to occur, insects must locate male plants to collect pollen and then carry it to a separate female plant in the population (Dötterl et al., 2014). Host location typically occurs through a combination of visual, olfactory, and reward cues. Salix species have a non-showy inflorescence arranged in a catkin form where male flowers are often yellow and female flowers tend to be green (Karrenberg et al., 2002; Füssel et al., 2007). Additionally, Salix species emit a complex mixture of volatile organic compounds that are important as olfactory signals to insects, and both male and female plants offer nectar rewards (Tollsten & Knudsen, 1992; Füssel et al., 2007). However, upon locating a host plant, some insects may rob flowers of their resources and not carry pollen between male and female individuals (Galen & Butchart, 2003).

Salix nigra, known by its common name black willow, is a tree-form, entomophilous willow that grows throughout the Eastern United States north to Maine, west to North Dakota and south to

Georgia (Burns & Honkala, 1990). The extensive range and productivity of S. nigra as well as its early bloom, typically February in its southern range through late June in more northern states, makes it an ideal resource for early emerging insects (Burns & Honkala, 1990; Ostaff et al.,

2015). Studying the mechanisms that S. nigra employs to attract insects as well as the influence of sex of tree on floral insect community through time is important in determining the competition for and potential stability of catkin resources, native oligolectic bee activity, and S. nigra reproductive success.

The goal of this study was to characterize the community of insects that visit S. nigra catkins and examine how the total floral community responded to tree sex, geographic position, volatile

114 organic compound (VOC) profiles, and secondary metabolites in catkins and leaves. For comparison, we also evaluated the community of floral insects captured using visual survey techniques and pan-traps placed in tree canopies. Finally, we examined the effect of tree sex,

VOCs, and survey year on the activity of three native Andrena bee species, including the willow oligolectic bees A. macoupinense and A. nigrae.

4.3 Methods

Population and site description

The target population of S. nigra was located in the West Virginia University Core Arboretum in

Morgantown, West Virginia (39.6462° N, 79.9811° W). The Core Arboretum is an old growth forest that contains 91 acres of native shrubs, trees, and herbaceous plants. It is located on a hillside that stretches between Monongahela Boulevard and the Monongahela River and contains riparian and floodplain sites with a small grove of S. nigra. The population contained thirty-two trees of which twelve were identified as female and twenty were male (Figure 1).

Visual survey technique

Visual surveys were performed in April through early May in years 2017, 2018, and 2019 on sunny days with minimal wind to help increase observation of small floral visitors. Visual observations were chosen as the survey method for this species due to the brittleness of the base of short shoots, which prevents the use of sweep nets (Beismann et al., 2000). To equalize observations among trees, survey branches were flagged containing approximately 300 catkin flowers for each individual tree. Trees were visually observed for 16 minutes per survey throughout their bloom (~ 2 weeks) and order of surveyed trees was randomized to account for

115

time of day for two female and four male trees. Insect specimens were carefully hand collected

from trees throughout the survey time for family/species level identification. Additionally,

survey month and day were recorded for each observation to use as covariates in models

analyzing individual insect activities as well as flower phenology. Finally, early season herbivore

activity, which was made up of observations of insects that feed on trees during and after catkins

were no longer present, was collected in 2019 at the end date of tree flowering by counting

number of herbivore occurrences on visual survey branches.

Canopy pan-trap technique

In the year 2019, pan-traps were constructed by painting three-ounce plastic cups with either

fluorescent blue, fluorescent yellow, or white paint (Guerra Paint and Pigment, New York).

Fluorescent paints were a mixture of 16 ounces of the fluorescent dispersion to 1 gallon of silica

flat paint. Three cups, one of each color, were affixed with velcro to a bucket lid as a platform to

be raised into tree canopies below flowering branches (Figure 2) on five female and six male

trees. A soapy trap solution was prepared by addition of approximately 5mL of organic

unscented dish soap into one gallon of water. Cups were filled ¾ of the way full with soap

solution. Traps were raised to the bottom of each tree canopies (2- 12 meters; average 4 meters)

at 9:00 am in the morning and emptied daily at 5:00 pm. Visual surveys were performed on the

same days. Captured insects from each cup were transferred to separate vials containing 70%

ethanol for later identification.

Insect identification

116

Abundant insects were identified to species level while rare insects were identified to family.

Native bee species identifications were validated by an expert (Sam Droege, Patuxent Wildlife

Research Center, personal communication). Vouchers of collected insect specimens were submitted to the West Virginia University Entomology Collection.

Flower volatile and tissue collection

Dormant branches were collected from the field in early spring for 6 female and 9 male Salix nigra trees along the river. Branches were allowed to root and flower in the Department of

Biology greenhouse in buckets of water. Volatile organic compounds (VOCs) were collected

using the dynamic headspace method (Keefover-Ring, 2013). An oven bag was placed over the

flowering branches and bags were secured with thin gauge wire around cotton pads that had been

wrapped near the base of the stems. At the top of each bag, polytetrafluoroethylene (PTFE) ports

were fixed and connected to chemical traps consisting of 65 mm long and 3 mm internal

diameter glass tubes packed with 20 mg of Super Q adsorbent (80/100 mesh size, DVB/

ethylvinylbenzene polymer) (Alltech Associates Inc., Deerfield, IL, USA). The traps were

connected to calibrated flow meters (Aalborg Instruments & Controls, Inc., Orangeburg, New

York, USA) and air was pulled through at a flow rate of 200 ml min-1 with an AirLite pump

(SKC Inc., Eighty Four, PA) modified to run with a 6 V battery. Controls consisted of oven bags

with an identical setup, but without an enclosed branch. Volatiles were collected for a three-hour

time period with flow meter maintaining a flow rate of 200 ml min-1. At the conclusion of each

sampling period, chemical traps were rinsed with 150 µl n-hexane (GC2, Honeywell Burdick &

Jackson, Morristown, NJ, USA) into GC vials with PTFE lined screw caps. All catkins were

counted in each bag, and maturation stage was noted. All catkins were subsequently lyophilized,

117

and the dry weight obtained of mature and immature (pre-anthesis) catkins separately. Prior to

GC-MS analysis, 40 µl of each sample was combined with 2 µl of an internal standard solution

(m-xylene in n-hexane). Collected floral volatile organic compounds (VOCs) were analyzed by gas chromatography (GC) with mass spectrometry (MS) detection.

VOC and metabolite characterization

Floral VOC samples were analyzed with a Thermo Trace 1310 GC coupled to a Thermo ISQ MS

with electron ionization (EI) at 70.0 eV at 250 °C, using helium as the carrier gas at 1.0 ml min−1

with the injector temperature set at 250 °C. Oven conditions included an initial temperature of 40

°C followed by an immediate ramp of 3 °C min−1 to 200 °C. Available standards, 1 μl of

samples, and a continuous series of n-alkanes (C8–C20; Sigma-Aldrich) were injected in the split mode onto a TR-5MS capillary column (30 m × 0.25 mm I.D., film thickness 0.25 μm; Thermo

Fisher Scientific). Compounds were identified with retention time matches to pure standards, mass spectra, and/or linear retention indexes calculated with the alkane series (Adams, 2001;

NIST, 2008; El-Sayed, 2013). Standard curves of available compounds were used to calculate final VOC results, which were expressed as ng compound g -1 DW hr -1.

Sample collection for chemical characterization

Catkins and leaves collected from 24 trees (14 males and 10 females) in the field and in the

greenhouse were characterized for five different metabolites; salicin, isosalicin, salicortin,

tremuloidin, and termulacin as well as total metabolites. Leaves were flash frozen in the field and

later shipped on dry ice to the University of Wisconsin-Madison, WI. The flash frozen catkins

and leaves were lyophilized, counted (catkins) and weighed (catkins and leaves), and then

118

ground with steel balls in plastic scintillation vials in a ball mill. Accurately weighed portions

(~15 mg) of powered leaf tissue were extracted with cold (4° C) methanol (1.00 ml) containing salicylic acid-d6 (Sigma-Aldrich, St. Louis, MO, USA) as an internal standard with sonication in an ice bath (15 min) and then centrifugation to obtain a clear supernatant for analysis.

Chemical analyses

2 μL of all standards and samples solutions were injected onto the UHPLC and separated peaks with a Waters Acquity CSH C-18 column (2.1 x 100 mm, 1.7 μm) at 40 °C with a flow rate of

0.5 mL min-1, using a gradient of water (solvent A) and acetonitrile (solvent B), both containing

0.1% formic acid. The PDA was configured to scan from 210-400 nm, with 1.2-nm resolution

and a sampling rate of 20 points·s-1. The MS operating parameters for were as follows: cone

potential, 30 V; capillary potential, 2500 V; extractor potential, 3 V; RF lens potential, 0.1 V;

source temperature, 120 °C; desolvation temperature, 250 °C; desolvation gas flow, 500 L h-1;

cone gas flow, 10 L h-1; infusion rate, 5 μL min-1; dwell time, 0.025 s.

Standard curves of methanol solutions, also containing the salicylic acid-d6 internal standard, of

various purified compounds were used to calculate the concentrations in the extracted leaves,

which were then normalized by sample dry weight and expressed as mg compound g-1.

Commercially available standards of salicin (Sigma-Aldrich), and salicortin, HCH-salicortin, tremuloidin, and tremulacin were used that had been previously isolated and purified from aspen foliage (Lindroth et al., 1986).

4.4 Statistical analysis

119

Floral insect community and sex effect on composition

To determine whether there were any sex differences in multivariate floral visitor community,

the R package vegan (Oksanen et al., 2019) was used ordinate the data using non-metric

multidimensional scaling (NMDS) using Bray-Curtis dissimilarities. Differences in floral visitor

communities among male and female flowers were determined using Analysis of Similarity

(ANOSIM), and significance was determined using 999 permutations to determine whether group assignments were significantly different from those generated by chance. Four survey types were tested to determine the stability of sex effects on floral community composition, including: 1) 2019 visual observations, 2) 2019 pan-traps, 3) 2019 total insect community across survey types and 4) visually observed communities through time from 2017-2019. Coordinates for dependent variables were extracted from the NMDS configuration and depicted in plots using a rescaled font to represent the three-dimensional projection. This was accomplished by rescaling all axis scores to a zero origin and scaling the font size relative to the product of the three rescaled axis scores.

Sex differences in tree chemistry

NMDS and ANOSIM were also utilized to determine if there were any sex differences in multivariate VOC and metabolite compositions by testing point grouping by sex of tree. Total monoterpenes, sesquiterpenes, VOC emissions, and metabolites were tested using a one-way

analysis of variance (ANOVA) in SAS software version 9.4 to determine if there were any

differences among sexes.

Tree chemistry and insect community relationship

120

Pairwise geographic distances were calculated among trees from GPS coordinates. VOC production per catkin was scaled to branch-level production using average catkin counts per branch for each individual tree to correlate to insect activity. Bray-Curtis dissimilarity matrices were generated for all datasets, including floral community, early season herbivore community, pairwise geographic distances, VOCs, and catkin/leaf metabolites. A Mantel test was utilized to determine whether differences in floral insect community were a function of pairwise geographic distances, or differences in catkin VOCs, catkin metabolites, or leaf metabolites. This analysis was also repeated for the tree early season herbivore community and catkin/leaf metabolite dissimilarities.

2019 survey method comparison and floral community

Twelve trees (5 female and 7 male) were surveyed in 2019 with visual and pan-trap techniques in the riparian and floodplain sites in the WVU Core Arboretum for a total of one-hundred and forty observation dataset. NMDS and ANOSIM were utilized to determine the effect of survey type on floral community composition. Visual observations were then merged with pan-trap capture counts for overlapping trees and dates for a final fifty-four observations. Independent numeric variables associated with survey day, including Julian date, military time, and temperature, were correlated with the NMDS configuration using the environmental fit vector analysis in vegan (envfit function). Variables that were found to significantly correlate to the community dataset were added as covariates in the nested analysis of covariance (ANCOVA) models.

121

Insects that were abundant in surveys, including A. macoupinense, A. morrisonella, Andrena nigrae, Lasioglossum coeruleum, and parasitic wasps belonging to the Braconidae family, were extracted from datasets. Additionally, species richness was calculated from the dataset as total number of species to visit each tree, and Shannon-Weaver diversity was calculated using the R package vegan. A nested ANCOVA was used to analyze differences in transformed insect counts, richness, and Shannon-Weaver diversity with the following model:

~ + + ( )& +

𝑦𝑦 𝑆𝑆𝑆𝑆𝑆𝑆 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑆𝑆𝑆𝑆𝑆𝑆 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 where () indicates the independent variable is nested in another variable.

Additionally, an environmental fit was conducted using the R package vegan with floral volatile compounds that were overlaid on the floral visitor community. The resulting patterns were then evaluated to select specific volatile compounds to look for linear relationships to insect activity using Pearson correlations with Bonferroni p-value corrections to account for multiple testing.

Insects chosen to test included Bombus sp., Andrena macoupinense & Andrena morrisonella,

Andrena nigrae, Miridae, Chalcosyrphus nemorum, and Sarcophagidae. VOCs of interest included acetophenone, cis-b-terpineol-2, ethyl-1-hexanol, germacrene-D, hexenyl-AC, octanal, octen-2-ol, and trans-3-pinanone.

Floral community through time

Among all three years, after accounting for mortality and branch loss, a total of six trees along the river overlapped among surveys, including two female and four male trees for 77

122

observation. NMDS and ANOSIM were utilized to determine the effect of survey year on floral

insect community composition. Independent numeric variables associated with survey day

including year, Julian date, military time, and temperature, and these variables were correlated with the NMDS configuration using the environmental fit vector analysis in vegan (envfit function). Variables that were found to significantly correlate to the community dataset were added as additional covariates in all nested ANCOVA models.

Insects that were abundant in surveys, including combined counts of A. macoupinense, A. morrisonella, and A. nigrae were extracted from the community dataset and a nested ANCOVA to analyze differences in transformed insect counts, richness, and Shannon-Weaver diversity with the following model:

~ + + ( )& +

𝑦𝑦 𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌 𝑆𝑆𝑆𝑆𝑆𝑆 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑆𝑆𝑆𝑆𝑆𝑆 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 where () indicates the independent variable is nested in another variable.

4.5 Results

Floral insect community and sex effect on composition

For visual surveys, across all three years of data, 3,160 insects were observed to visit flowers. Of those observations, 88.9% were hymenopteran, 9.6% were dipteran, 1.3% were hemipteran, and

0.2% were coleopteran. Additionally, of all insects observed, bees belonging to the genus

Andrena made up 69.4% of the visually surveyed community. NMDS indicated that that the appropriate number of dimensions for 2019 visual surveys, 2019 pan-traps, 2019 total

123

community (Figure 3), and 2017-2019 community (Figure 4) was four (stress = 0.11), three

(stress = 0.10), three (stress = 0.12), and four (stress = 0.12), respectively. An ANOSIM indicated that the floral community composition was dependent upon the sex of the tree for all

survey types, 2019 total community, and across years (R > 0.1000, p-value < 0.05; Table 1).

Sex differences in tree chemistry

The NMDS of catkin VOCs indicated the appropriate number of dimensions for analysis was three (stress = 0.10) while the number of dimensions for leaf metabolites and catkin metabolites was two (stress = 0.10). An ANOSIM (Table 2) indicated that the catkin VOC composition and leaf metabolite composition were not significantly different between male and female trees (R =

-0.1191, p-value = 0.914; R = -0.00471, p-value = 0.449). However, the catkin metabolite composition was different between the sexes (R = 0.1787, p-value = 0.027; Figure 5). The total amount of catkin and leaf metabolites did not differ between the sexes (catkins, one-way

ANOVA F1,28 = 2.741, p-value = 0.109; leaves, F1,30 = 1.549, p-value = 0.2229). Furthermore,

total monoterpenes (F1,13 = 0.02755, p-value = 0.8707), sesquiterpenes (F1,13 = 0.2623, p-value =

0.6171), and VOC emissions (F1,13 = 0.004865, p-value = 0.9455) were not significantly

different between the sexes (Figure 6).

Tree chemistry and insect community relationship

Mantel tests (Table 3) indicated that there was no relationship between pairwise geographic

distances and floral visitor community (rm = 0.026, p-value = 0.450). Similarly, distances

between catkin volatile composition, leaf metabolite composition, and catkin metabolite composition were not related to differences in floral visitor community (rm = -0.200, p-value =

124

0.100; rm = 0.002, p-value = 0.480; rm = 0.062, p-value = 0.380, respectively). There was a significant positive relationship between differences in catkin metabolite composition and early season herbivore community (rm = 0.416, p-value = 0.010).

2019 survey method comparison and floral community

An NMDS with results from both survey methods indicated the appropriate number of dimensions for analysis was three (stress = 0.12; Figure 7a). An ANOSIM indicated that the floral community composition was dependent upon survey method (R = 0.6656, p-value =

0.001). Pan-traps captured 284 total insects with the majority of insects captured belonging to the orders Diptera (33%) and Coleoptera (41%). There were 1,531 insect observations made during visual surveys with the majority (87%) belonging to the order Hymenoptera (Figure 7b).

The NMDS vector analysis indicated that only Julian date was significantly correlated with the

NMDS configuration (Vector Max R = 0.3036; p-value = 0.001; Table 4). Julian date was then selected for use as a covariate in all nested ANCOVA models. A nested ANCOVA revealed that the occurrence of A. nigrae (F13,40 = 5.6233; p-value = 0.0310), A. morrisonella (F13,40 = 7.2722, p-value = 0.0140), and L. coeruleum (F13,40 = 4.2768, p-value = 0.0500) was dependent on the sex of the tree (Table 5,7; Figure 7). Additionally, species richness (F13,40 = 17.749; p-value =

0.0004) and Shannon-Weaver diversity (F13,40 = 28.936; p-value < 0.0001) also differed between the sexes, with males demonstrating higher values (Table 6,7; Figure 8). Finally, the abundance of A. macoupinense differed significantly among trees (F13,40 = 3.3835, p-value = 0.0030).

125

A significant correlation was found between the abundance of A. nigrae and two volatile

compounds: acetophenone (ρ = 0.9122, p-value = 0.05) and octen-2-ol (ρ = 0.9393, p-value =

0.01) (Figure 9). Additional VOCs and insect activities were not significantly correlated (p-value

> 0.05).

Floral community through time

An ANOSIM indicated that the floral community composition differed among years (R =

0.1669, p-value = 0.001). The vector analysis indicated that only Julian date was significantly

correlated with the NMDS dimensions (Vector Max R = 0.0802; p-value = 0.047; Table 8). A

nested ANCOVA (Table 9,10; Figure 10) showed that average species richness (F8,68 = 32.2841,

p-value = 0.0046) and Shannon-Weaver diversity (F8,68 = 34.6569; p-value = 0.0041) were

dependent on the sex of the tree. Additionally, flowering time occurred substantially earlier in

2017 than in the other survey years (Figure 11a), which may be related to the dependence of A.

nigrae activity on the survey year (F8,68 = 5.1049; p-value = 0.0086; Figure 11b).

4.6 Discussion

The total floral community composition was strongly influenced by the sex of tree across survey

techniques and in all years. Male trees consistently attracted a more diverse and unique insect

assemblage on their flowers when compared to female trees. However, there was no relationship

detected between the sex of the tree and the floral VOC composition, indicating that the scent of

male and female flowers among trees was not different. Total emission of monoterpenes,

sesquiterpenes, and all VOCs were also found to be similar among sexes. Differences in floral

VOCs and plant tissue defenses among trees were also not strong drivers of the community of

126

insects that visited the flowers. Thus, we found no evidence that the influence that sex of tree on

floral insect assemblages was due to the effects of VOCs or metabolite composition, although

more subtle relationships might be revealed with more intensive sampling.

Our results are unusual given that dioecious plants, in which both sexes offer rewards, are often

dimorphic in both scent composition and emission levels (Füssel, 2007; Ashman, 2009; Okamoto

et al., 2013). Nevertheless, this pattern does not always hold true in the Salix genus. For

example, S. repens, S. bicolor, S. caprea, and S. cinerea all have similar overall volatile

composition among male and female flowers (Tollsten & Knudsen, 1992; Füssel, 2007). This may reflect differences in the balance between visual and olfactory cues among Salix species, although this hypothesis remains to be robustly tested.

Although male and female trees did not differ in VOC and leaf metabolite composition, there was a sex effect on the metabolite composition of catkins. We also found that catkin metabolites exerted a strong influence over herbivore composition, suggesting that floral defenses are utilized differently among male and female trees, in turn attracting unique assemblages of herbivores. In the early season, trees may be at risk of herbivores feeding directly on flowers which act as a large resource sink in plant tissues (Wäckers et al., 2007). Floral larceny is a threat to female reproduction since trees must maintain flowers through seed production, and this relationship may be antagonized in a dioecious system where chance of accidental pollination is rare (Maloof & Inouye, 2000; Richardson, 2004). Thus, it is important that females invest more

resources toward defense of catkins, as supported by our finding that female individuals emitted

more (3E)-hexenyl acetate and (2E)-hexenyl acetate. Hexenyl acetates have been frequently

127

characterized as common green leaf volatiles emitted upon the crushing of plant tissue (Whitman

& Eller, 1990; Heil et al., 2008; Wei & Kang, 2011) and they appear to also be a common

component in floral scent (Tollsten & Knudsen, 1992; Kaiser, 1994; Messinger, 2006). Hexenyl

acetates are associated with the attraction of insect parasites, which may provide female flowers

in S. nigra added protection benefits against floral resource theft (Whitman & Eller, 1990;

Ngumbi et al., 2009).

In our 2019 floral community surveys, pan-traps were employed to test differences in survey

techniques and characterize total insect community shifts among male and female trees. We

determined that pan-trapping and visual survey techniques captured vastly different samples of

the floral visitor community. Pan-traps were very effective at capturing small insects which

belonged to Diptera and Coleoptera, explaining much of the disparity between techniques.

However, hymenopterans made up 87% of insects counted during visual surveys with 67% of bee observations coming from native Andrena spp. Conversely, hymenopterans only made up

22% of insects captured in pan-traps with only 8% of the bees captured belonging to Andrena

spp. Unscented pan-traps filter the insect community, which was presumably attracted to the

flowers on the tree, to those relying heavily on visual cues (Laubertie et al., 2006; Tuell &

Isaacs, 2009; Vrdoljak & Samways, 2012). The lack of native Andrena spp. attraction to pan-

traps may indicate that both visual and volatile cues are important in their association rather than

only visual cues. Additionally, given that the pan-trap data increased the diversity disparity

observed between male and female flowers in 2019, it appears that the initial visual component

of male pollen may be influencing the differences in the larger floral community shift observed

128

among male and female trees. This is further supported by the differential rates of capture of the

hemipterans and hymenopterans in yellow cups and coleopterans in blue cups.

Surveys completed in 2019 revealed the sex of tree was important for many of the

Hymenopteran distributions throughout our site. A. morisonella, L. coeruleum, and a parasitoid

wasp species in the family Braconidae all showed higher activity on male trees indicating a

prioritization of male rewards. A. macoupinense was equally observed among male and female

trees. Interestingly, A. nigrae was more actively visiting female flowers on trees in the study site.

Given the non-showy nature of the Salix female catkins, this suggests that the species may be

more tightly coupled to the volatile rather than visual cues in order to locate an appropriate host.

In support of this, A. nigrae was the only insect whose activity was significantly correlated to level of specific floral volatiles. The number of visits of A. nigrae increased with increasing levels of both acetophenone and octen-2-ol.

Despite detection of sex differences in activity of native bees in 2019, Andrena spp. were stable

among male and female flowers through time, indicating that all three species are important in

contributing to sexual reproduction of S. nigra. Additionally, the total activity of A.

macoupinense and A. morisonella were stable across years, but A. nigrae had far fewer catkin

visits in 2017 when compared to later years. Adult emergence of A. macoupinense has been

recorded mid-March through May, A. nigrae is often found April through May, and A.

morisonella is frequent from May to early June (Stevens, 1949; Ribble, 1968, 1974). Of the three species, A. nigrae emergence is more closely coupled to the bloom of local S. nigra trees.

Additionally A. nigrae has been recorded as a primary pollinator of S. nigra in two other states,

129

Illinois and North Dakota, indicating that it may be more tightly coupled to the biology of S. nigra than other observed native bees in our site (Stevens, 1949; Ribble, 1968).

When examining the phenology of flowers through time, the bloom time of trees in 2017 occurred earlier than subsequent years due to uncharacteristically warm weather in early March followed by much cooler temperatures through April and early May. The abnormality of temperature may have decoupled the availability of floral resources from local A. nigrae population emergence. Additionally, cooler temperatures may have had a negative effect on volatilization of catkin scent, leading to the inability of native bees to effectively locate their host. The lack of activity difference detected for A. macoupinense and A. morisonella may indicate that local populations effectively supplement their diets with resources from additional plant species, such as other local Salix, including S. discolor and S. bebbiana, or Prunus and

Amelanchier spp., both of which are native to West Virginia and listed as occasional hosts

(Ribble, 1974).

While male trees attract a more diverse community of floral visitors, males also overproduce resources (average pollen:ovule ratio ~1500:1, J. Grady, unpublished data), thus minimizing competition between native pollinators and pollen/nectar thieves. Conversely, female individuals lack strong visual cues and, despite having similar floral volatile composition, appear to have different protective chemistry composition in their catkins allowing them to preserve their rewards directly for the cross-pollinating Andrena species. Competition for resources does not seem to be a direct threat to Andrena activity or successful sexual reproduction of S. nigra.

However, we show that one native bee species may be vulnerable when local tree bloom time is

130 decoupled from their seasonal emergence. For most native pollinators and plants, partial asynchrony with bloom time is not a threat if additional hosts are present and insect phenology has time to adjust to permanently altered conditions such as those caused by climate change

(Willmer, 2012; Bartomeus et al., 2013). For oligolectic species like A. nigrae that rarely collect from hosts outside of a single plant group or species, even mild decoupling could have drastic implications for locally adapted bee populations. As a result, plant sexual reproduction will be negatively impacted, especially when the cross-pollinating community is made up of specialists, and accidental pollination becomes rare.

Overall, differences in floral community composition and diversity among male and female trees of S. nigra appears to be strongly driven by visual cues. However, the main cross pollinators,

Andrena spp., rely on both visual and specific volatile cues to locate male individuals and carry pollen to the less showy female trees. Additionally, the interannual variability in flowering time highlights the potential effects of climate on pollinator activity and tree sexual productivity.

Inter-annual variability in the S. nigra – Andrena interaction illustrates that shifting seasonal transitions could detrimentally affect plants that depend on early-emerging arthropods for sexual reproduction as well as the arthropods that depend on resources provided by early-flowering plants.

131

4.7 Literature cited

Adams RP. 2001. Identification of essential oil components by gas chromatography/mass

spectrometry. Carol Stream, IL: Allured Publishing Corporation.

Argus GW. 2011. An experimental study of hybridization and pollination in Salix (willow).

Canadian Journal of Botany 52: 1613–1619.

Ashman TL. 2009. Sniffing out patterns of sexual dimorphism in floral scent. Functional

Ecology 23: 852–862.

Bartomeus I, Park MG, Gibbs J, Danforth BN, Lakso AN, Winfree R. 2013. Biodiversity

ensures plant-pollinator phenological synchrony against climate change (M Eubanks,

Ed.). Ecology Letters 16: 1331–1338.

Beismann H, Wilhelmi H, Baillères H, Spatz H, Bogenrieder A, Speck T. 2000. Brittleness of

twig bases in the genus Salix: fracture mechanics and ecological relevance. Journal of

Experimental Botany 51: 617–633.

Burns RM, Honkala BH. 1990. Silvics of North America: Volume II: hardwoods. Washington,

D.C., USA: United States Department of Agriculture (USDA), Forest Service.

Danforth B. 2007. Bees. Current Biology 17: 156–161.

Dötterl S, Glück U, Jürgens A, Woodring J, Aas G. 2014. Floral reward, advertisement and

attractiveness to honey bees in dioecious Salix caprea (AG Dyer, Ed.). PLoS ONE 9:

e93421.

El-Sayed AM. 2013. The pherobase: database of insect pheromones and semiochemicals.

Füssel U. 2007. Floral scent in Salix L. and the role of olfactory and visual cues for pollinator

attraction of Salix caprea L. [PhD-Thesis].

Füssel U, Dötterl S, Jürgens A, Aas G. 2007. Inter- and intraspecific variation in floral scent in

132

the genus Salix and its implication for pollination. Journal of Chemical Ecology 33: 749–

765.

Galen C, Butchart B. 2003. Ants in your plants: effects of nectar-thieves on pollen fertility and

seed-siring capacity in the alpine wildflower, Polemonium viscosum. Oikos 101: 521–

528.

Heil M, Lion U, Boland W. 2008. Defense-inducing volatiles: in search of the active motif.

Journal of Chemical Ecology 34: 601–604.

Kaiser R. 1994. Trapping, investigation and reconstitution of flower scents. Springer

Netherlands.

Karrenberg S, Edwards P, Kollmann J. 2002. The life history of Salicaceae living in the

active zone of floodplains. Freshwater Biology 47: 733–748.

Keefover-Ring K. 2013. Making scents of defense: do fecal shields and herbivore-caused

volatiles match host plant chemical profiles? Chemoecology 23: 1–11.

Laubertie EA, Wratten SD, Sedcole JR. 2006. The role of odour and visual cues in the pan-

trap catching of hoverflies (Diptera: Syrphidae). Annals of Applied Biology 148: 173–

178.

Lindroth RL, Scriber JM, Hsia MTS. 1986. Differential responses of tiger swallowtail

subspecies to secondary metabolites from tulip tree and quaking aspen. Oecologia 70:

13–19.

Linsley EG. 1958. The ecology of solitary bees. Hilgardia 27: 543–599.

Maloof JE, Inouye DW. 2000. Are nectar robbers cheaters or mutualists? Ecology 81: 2651–

2661.

Messinger OJ. 2006. The role of visual and olfactory cues in host recognition for the specialist

133

bee genus Diadasia, and implications for the evolution of host choice [PhD Thesis].

Ngumbi E, Chen L, Fadamiro HY. 2009. Comparative GC-ead responses of a specialist

(Microplitis croceipes) and a generalist (Cotesia marginiventris) parasitoid to cotton

volatiles induced by two caterpillar species. Journal of Chemical Ecology 35: 1009–

1020.

NIST. 2008. National institute of standards and technology mass spectral library. Washington,

D.C.: National Institute of Standards and Technology, US Department of Commerce.

Okamoto T, Kawakita A, Goto R, Svensson GP, Kato M. 2013. Active pollination favours

sexual dimorphism in floral scent. Proceedings of the Royal Society B: Biological

Sciences 280.

Oksanen J, Blanchet FG, Friendly M, Kindt R, Legendre P, McGlinn D, Minchin PR,

O’Hara RB, Simpson GL, Solymos P, et al. 2019. vegan: community ecology package.

R package version 2.5-4. https://cran.r-project.org/package=vegan.

Ostaff DP, Mosseler A, Johns RC, Javorek S, Klymko J, Ascher JS, Johns A, Javorek RC,

And Ascher J. 2015. Willows (Salix spp.) as pollen and nectar sources for sustaining

fruit and berry pollinating insects. Canadian Journal of Plant Science 95: 505–516.

Ribble DW. 1968. Revisions of two subgenera of Andrena: Micrandrena (Ashmead) and

Derandrena. New subgenus (Hymenoptera: Apoidea). Bulletin University of Nebraska

Museum 8: 237–394.

Ribble DW. 1974. A revision of the bees of the genus Andrena of the western hemisphere

subgenus Scaphandrena. American Entomological Society 100: 101–189.

Richardson SC. 2004. Are nectar-robbers mutualists or antagonists? Oecologia 139: 246–254.

Stevens OA. 1949. Native bees. Experimental Station Bimonthly Bulletin 12: 90–98.

134

Straka J, Rina Cern K, Mach L, Zemenov M, Keil P. 2014. Life span in the wild: the role of

activity and climate in natural populations of bees. Functional Ecology 28: 1235–1244.

Tamura S, Kudo G. 2000. Wind pollination and insect pollination of two temperate willow

species, Salix miyabeana and Salix sachalinensis. Plant Ecology 147: 185–192.

Tollsten L, Knudsen JT. 1992. Floral scent in dioecious Salix (Salicaceae)- a cue determining

pollination system? Plant Systematics and Evolution 182: 229–237.

Tuell JK, Isaacs R. 2009. Elevated pan traps to monitor bees in flowering crop canopies.

Entomologia Experimentalis et Applicata 131: 93–98.

Vrdoljak SM, Samways MJ. 2012. Optimising coloured pan traps to survey flower visiting

insects. Journal of Insect Conservation 16: 345–354.

Wäckers FL, Romeis J, van Rijn P. 2007. Nectar and pollen feeding by insect herbivores and

implications for multitrophic interactions. Annual Review of Entomology 52: 301–323.

Wei J, Kang L. 2011. Roles of (Z)-3-hexenol in plant-insect interactions. Plant Signaling and

Behavior 6: 369–371.

Whitman DW, Eller FJ. 1990. Parasitic wasps orient to green leaf volatiles. Chemoecology 1:

69–76.

Willmer P. 2012. Ecology: pollinator–plant synchrony tested by climate change. Current

Biology 22: R131–R132.

135

4.8 Tables and Figures

Table 1. NMDS and ANOSIM (Bray-Curtis dissimilarity) results for multivariate floral visitor

composition tested against sex grouping (male vs female). p-values < 0.05 (bolded) indicate that floral visitor composition is more similar within replicate observations of sex group rather than among all observations.

Survey # dimensions Stress ANOSIM R p-value

2019 visual 4 0.11 0.3110 0.001

2019 pan-traps 3 0.10 0.1532 0.017

2019 total community 3 0.12 0.3077 0.001

2017-2019 community 4 0.12 0.1018 0.032

136

Table 2. NMDS and ANOSIM (Bray-Curtis dissimilarity) results for multivariate chemistry

composition tested against sex grouping (male vs female). p-values < 0.05 (bolded) indicate that chemistry composition is more similar within replicate observations of sex group rather than among all observations.

Multivariate # dimensions Stress ANOSIM R p-value

response

VOCs 3 0.10 -0.1191 0.914

Catkin metabolites 2 0.10 0.1787 0.027

Leaf metabolites 2 0.10 -0.00471 0.449

137

Table 3. Mantel test results comparing pairwise Bray-Curtis dissimilarity matrices among insect communities and chemistry composition of flowers and leaves. Bolded p-values and positive rm indicate a significant test, suggesting that similarity in chemistry composition relates to similarity in insect community assemblage.

Insect matrix Chemistry matrix rm p-value

Floral community Catkin VOCs -0.300 0.07

Floral community Catkin metabolites 0.062 0.380

Floral community Leaf metabolites 0.002 0.480

Herbivore community Catkin metabolites 0.416 0.010

Herbivore community Leaf metabolites 0.275 0.07

138

Table 4. Environmental fit results for 2019 floral community NMDS analysis. Bolded p-values

(<0.05) indicate a significant correlation of the independent variable with the NMDS configuration.

Independent variable Vector Max R p-value

Julian date 0.3036 0.001

Temperature 0.0582 0.241

Military time 0.0447 0.308

139

Table 5. Nested ANCOVA model results for 2019 analysis of individual native bee abundances.

Bolded p-values (<0.05) indicate significant model effects. y ~ Sex + Location + Tree(Sex)&Random + Julian date

Dependent variable: Andrena nigrae Source Degrees of Sum of Mean F- p-value freedom squares square Ratio

Model 13 43.48185 3.34476 2.4947 0.0134

Error 40 53.62887 1.34072

C. Total 53 97.11072 Dependent variable: Andrena macoupinense Source Degrees of Sum of Mean F- p-value freedom squares square Ratio

Model 13 5.604838 0.431141 3.9674 0.0004

Error 40 4.34681 0.10867

C. Total 53 9.951648 Dependent variable: Andrena morrisonella Source Degrees of Sum of Mean F- p-value freedom squares square Ratio

Model 13 4.347317 0.334409 2.9616 0.0042

Error 40 4.516573 0.112914

C. Total 53 8.86389 Dependent variable: Lasioglossum coeruleum Source Degrees of Sum of Mean F- p-value freedom squares square Ratio

Model 13 3.101719 0.238594 2.2816 0.023

Error 40 4.182974 0.104574

C. Total 53 7.284693

140

Table 6. Nested ANCOVA model results for 2019 analysis of calculated community metrics.

Bolded p-values (<0.05) indicate significant model effects. y ~ Sex + Location + Tree(Sex)&Random + Julian date

Dependent variable: Species richness

Source Degrees of Sum of Mean F- p-value freedom squares square Ratio

Model 13 307.5722 23.6594 3.3871 0.0015

Error 40 279.4093 6.9852

C. Total 53 586.9815

Dependent variable: Shannon-Weaver diversity

Source Degrees of Sum of Mean F- p-value freedom squares square Ratio

Model 13 9.700905 0.746223 5.2588 <0.0001

Error 40 5.675971 0.141899

C. Total 53 15.37688

141

Table 7. Test of random effects p-values extracted from nested ANCOVA for most abundant floral visitors as well as calculated species richness and Shannon Weaver diversity for 2019 survey analysis. Bolded values indicate that the independent variable had a significant effect on the dependent variable (p-value < 0.05).

Andrena Andrena Andrena Lasioglossum Species Shannon- nigrae macoupinense morrisonella coeruleum richness Weaver diversity Sex 0.031 0.5447 0.014 0.0500 0.0004 <0.0001

Location 0.5645 0.3587 0.8326 0.9577 0.9536 0.7348

Tree 0.2046 0.003 0.4729 0.1635 0.5944 0.6917

Julian 0.5645 0.6261 0.0600 0.6141 0.5800 0.2839 date

142

Table 8. Environmental fit results for floral community NMDS analysis across years (2017-

2019). Bolded p-values (<0.05) indicate a significant correlation of the independent variable with the NMDS configuration.

Independent variable Vector Max R p-value

Year 0.0802 0.047

Julian date 0.0815 0.046

Temperature 0.0491 0.18

143

Table 9. Nested ANCOVA model results for year analysis. Bolded p-values (<0.05) indicate significant model effects. y ~ Sex + Year + Tree(Sex)&Random + Julian date

Dependent variable: Andrena macoupinense & Andrena morrisonella

Source Degrees of Sum of Mean F- p-value freedom squares square Ratio

Model 8 27.17312 3.39664 1.6702 0.1218

Error 68 138.2903 2.03368

C. Total 76 165.4635

Dependent variable: Andrena nigrae Source Degrees of Sum of Mean F- p-value freedom squares square Ratio

Model 8 46.10628 5.76328 4.9103 <0.0001

Error 68 79.81222 1.17371

C. Total 76 125.9185 Dependent variable: Species richness Source Degrees of Sum of Mean F- p-value freedom squares square Ratio

Model 8 82.47375 10.3092 4.8422 <0.0001

Error 68 144.773 2.129

C. Total 76 227.2468 Dependent variable: Shannon-Weaver diversity Source Degrees of Sum of Mean F- p-value freedom squares square Ratio

Model 8 5.76738 0.720923 7.3942 <0.0001

Error 68 6.62987 0.097498

C. Total 76 12.39725

144

Table 10. Test of random effects p-values extracted from nested ANCOVA for most abundant floral visitors as well as calculated species richness and Shannon Weaver diversity. Bolded values indicate that the independent variable had a significant effect on the dependent variable

(p-value < 0.05).

Andrena nigrae Species richness Shannon-Weaver diversity

Sex 0.1475 0.0046 0.0041

Year 0.0086 0.7303 0.9019

Tree 0.1134 0.4356 0.2488

Julian date 0.2171 0.2467 0.2617

145

Figure 1. Map of all thirty Salix nigra tree locations in WVU Core Arboretum. Pin shapes

indicate tree selected for floral insect community survey. Points indicate additional individuals in

the Arboretum.

Figure 2. Tree canopy pan-trap design. Camouflaged buckets were added to hang below river

traps to add weight, preventing winds from jostling traps.

Figure 3. Non-metric multidimensional plot (dimensions = 3; stress = 0.12) of insect floral

community with groupings indicated by color for sex of tree (ANOSIM R =0.3077, p-value =

0.001) for 2019 analysis. Size of species text indicates position relative to the dimensional space,

for example, larger font indicates forward projection along the positive values of all axes.

Figure 4. Non-metric multidimensional plot (dimensions = 4; stress = 0.12) of 2017-2019 insect

floral community with groupings indicated by color for sex of tree (ANOSIM R = 0.1232, p- value = 0.011). Size of species text indicates position relative to the dimensional space, for example, larger font indicates forward projection along the positive values of all axes.

Figure 5. Total concentration averages of catkin metabolites, leaf metabolites and catkin VOCs for male and female trees. Letters to the left of boxes indicate significantly different means (p- value < 0.05) determined by one-way ANOVA test.

146

Figure 6. Total concentration averages of monoterpenes and sesquiterpenes VOCs for male and female trees. Letters to the left of boxes indicate significantly different means (p-value < 0.05) determined by one-way ANOVA test.

Figure 7. (a) Non-metric multidimensional plot (dimensions = 3; stress = 0.12) of insect floral community with groupings indicated by color for survey method (ANOSIM R = 0.6656, p-value

= 0.001). (b) Breakdown of insect capture/observation for two canopy survey methods.

Additional pie charts around pan-trap pie chart indicate the percentage of that order captured in different colors of pan-traps.

Figure 8. Average activity of most common floral visitors and calculated community metrics from 2019 S. nigra surveys. Letters to the left of boxes indicate significantly different means as determined by a Tukey’s HSD (p-value < 0.05).

Figure 9. Correlation plots of A. nigrae with VOC compounds.

Figure 10. Average values of species richness and Shannon-Weaver diversity for female and male trees from year analysis model. Letters to the left of boxes indicate significantly different means as determined by a Tukey’s HSD (p-value < 0.05) for each separate nested ANCOVA model.

147

Figure 11. (a) Amount of time and date in each year (2017, 2018, and 2019) in which catkins

were actively attractive to insects. End date for male individuals indicates trees have dropped all

catkins while female trees no longer have receptive stigmas (brown and shriveled) or active

insect in canopies. (b) Average count of A. nigrae visits to S. nigra catkins in survey years 2017,

2018 and 2019 from year analysis model. Letters to the left of boxes indicate significantly different means as determined by a Tukey’s HSD (p-value < 0.05).

148

Figure 1.

149

Figure 2.

150

Figure 3.

151

Figure 4.

152

Figure 5.

153

Figure 6.

154

Figure 7.

155

Figure 8.

156

Figure 9.

157

Figure 10.

158

Figure 11.

159

Chapter 5: Conclusions

160

5.1 Overview of research goals

For my dissertation, I used Populus and Salix as models in order to study the host genetic

influence on biotic communities by: 1) utilizing a hybrid backcross family to identify loci

underlying fungal and insect associations in Populus, and genomic comparisons of P.

trichocarpa and P. deltoides to understand genetic mechanisms underlying host plant-biotic

relationships; 2) using an association population of P. trichocarpa to identify and uncover

potential biological function of candidate genes underlying individual arthropod associations and

entire community composition; and 3) studying the impacts of dioecy on floral insect

communities using insect pollinated S. nigra.

5.2 Hybrid mediation of biotic associations

In Chapter 2, I identified several fungal pathogens and insect species that were segregating

among our P. deltoides x P. trichocarpa pseudo-backcross family progeny. QTL analysis

revealed a combined 7 loci controlling the distributions of 5 surveyed species, including

Melampsora sp. fungal leaf rust, S. musiva fungal leaf spot, S. musiva fungal stem canker, M.

vagabunda branch galling aphid, and Phyllocolpa sp. leaf folding sawfly. An overlapping

genetic interval on Chr04 was discovered to be associated with the biotrophic Melampsora sp.

and necrotrophic S. musiva fungus. Upon further investigation I found that the hybrids were

mediating the interaction between the two pathogens, resulting in competitive exclusion (Orton

& Brown, 2016), which led to an inflation of the relationship between the Chr04 interval and

occurrence of S. musiva leaf spot symptoms (Koskella et al., 2006; Abdullah et al., 2017). An additional QTL interval was also discovered to co-occur with the Phyllocolpa sp. Chr10 association, which was important in the host-plant production of the metabolite gentisyl alcohol

161

5-O-glucoside. Given that the positive allele contributing to an increase in both phenotypes came from P. trichocarpa, the metabolite may be acting as an ovipositional cue for egg-laying female

sawflies.

Genomic comparisons of the parents of the backcross family revealed several interesting

observations. The Chr16 interval that was important in the severity of S. musiva canker

symptoms did not overlap with the four loci that have previously been implicated in P.

trichocarpa susceptibility (Muchero et al., 2018), which suggests that the hybrids are mediating

the pathogen interactions using a potentially analogous, but distinct, mechanism. The QTL interval contained a tandem repeat of G-type lectin receptor-like kinases that were expanded in the P. deltoides parent, which may play a similar role to the receptor-like kinase family known to be important in controlling the infection severity of the pathogen in the P. trichocarpa study

(Muchero et al., 2018). Our other QTL intervals revealed functional enrichments and multiple recent tandem duplications of genes that are important in resistance to biotic stress such as cell-

wall loosening, cell death signaling, secondary metabolite production, and R-genes as well as

ones that may be important in susceptibility such as sugar transportation that can make source

tissues into nutritive sinks. Finally, I found that recent tandem gene expansion was enriched in the P. delotides parent genome, but not in the P. trichocarpa parent. Most of the herbivores and pathogens surveyed in our study have been co-evolving with P. deltoides, leading to a longer history of gene-for-gene interactions that potentially left a lasting impact on the genome structure in the form of recent tandem gene duplications.

162

5.3 P. trichocarpa community genetics

In Chapter 3, I used an association population of P. trichocarpa and surveys of arthropod activity

across three common gardens and two years to better understand the genetic control of abundant

arthropods and community composition. I discovered that, despite the arthropod species

distributions being highly variable among each garden, there was a strong host-genetic influence mediating the assembly of herbivores and predators in all gardens and years. Furthermore, the occurrence of abundant arthropods and community metrics, such as arthropod richness (total

number of unique species on plant tissue), abundance (total number of arthropods on plant

tissue), and Shannon Weaver diversity, were found to be strongly heritable across all sites. Two

gardens, Clatskanie 2012 and Corvallis 2015, had an adequate number of genotypes surveyed for

genome wide association analysis (GWAS). Single variable traits including abundance of

arthropods such as Harmandia sp., Phyllocolpa sp., and binarized occurrences of Chrysomela

scripta, Phyllonorycter sp. blotch miner, Phyllocnistis populiella serpentine miner, and

Corythucha sp. lace bug. Additionally, community metrics calculated from each garden, such as

richness, abundance, and diversity, were included in single-trait analyses. Finally, a multi-trait

GWAS was also implemented for the axis scores extracted from the NMDS configurations for

the Clatskanie and Corvallis sites to look for genes important in underlying arthropod

community composition.

Significantly associated SNPs were identified for 10 of the 14 traits on or near 88 potential

candidate genes of interest and, of these, only 32 yielded significant network results based on

genes that were co-expressed with candidate genes, methylation data for multiple P. trichocarpa

tissue types (Vining et al., 2012), and GWAS for pyMBMS and metabolite profiles

163

(Tschaplinski et al., 2012). Candidate genes with biologically meaningful functions were

selected for further investigation from networks based on a depth (number of connected layers)

cutoff of 3 and a breadth (number of neighbor genes) cutoff of 500, leaving us with 6 genes of

interest associated with the distribution of Phyllonorycter sp. blotch miner in the Clatskanie

garden and arthropod richness and composition from the Corvallis garden. Candidate genes

important in the association of the specialist herbivore Phyllonorycter sp. were found to be

important in tissue senescence and metabolite production, whereas genes important in arthropod

richness and community composition had functions that were capable of more broadly targeting

multiple species of arthropods, such as terpenoid production, RNA interference, and

transmembrane signaling.

5.4 S. nigra floral community

In Chapter 4, I was interested in exploring the effect of plant dioecy on communities of insects in

a natural population of S. nigra. To achieve this, I surveyed flowers for insect visitors using both visual and pan trapping techniques across three years, insect herbivores that feed on trees during flowering and after catkins are no longer present. Various traits collected from the S. nigra trees, including sex of the individual, catkin volatile organic compounds (VOCs), catkin metabolites, and leaf metabolites, were explored as potential candidates mediating the interactions of the trees with their arthropod communities. Total composition of floral visitors did not appear to be related to defensive or volatile chemistry composition of flowers or leaves but were strongly driven by sex of tree, indicating that visual cues, such as the yellow color of male pollen are important in driving the diversity disparity observed among male and female trees. Composition

164

of herbivores however, did appear to be related to tree defensive metabolites, especially in the

case of catkin phenolics, which were also strongly driven by sex of tree.

I identified three insects that frequently visited both male and female trees that are the most

likely cross-pollinators of S. nigra in my site. All three were identified as species of the solitary

ground nesting bees Andrena, including A. morisonella, A. macoupinense, and A. nigrae. Two of

the species, A. morisonella and A. macoupinense, showed higher activity on male catkins, while

A. nigrae showed stronger preference for female trees. Female catkins lack attractive visual cues,

suggesting that insects that can easily locate their resources may be relying on specific volatile

cues. This was supported by my finding that A. nigrae activity was strongly correlated to the

levels of two volatile compounds, including acetophenone and octen-2-ol. My final finding in

this chapter was that, through time, the activity of A. morisonella and A. macoupinense stayed

stable on S. nigra flowers, but the activity of A. nigrae was lower in 2017 when flowers bloomed

the earliest. This created an asynchrony between the emergence of the native bees and

availability of floral resources which is not only a threat to the sexual reproduction of S. nigra

trees, but to the survival of local A. nigrae populations that rely on this single host species. Host-

insect asynchrony in tightly-linked relationships, such as the Salix-Andrena system, threaten to

become more frequent and widespread given global impacts of climate change altering their ecosystems (Willmer, 2012; Bartomeus et al., 2013).

5.5 Overall conclusions

165

In conclusion, I found that hybrids can be a valuable tool for uncovering loci and genomic

patterns associated with biotic distributions. However, hybrid mediations of biotic interactions

can also lead to unique genetic associations that would not be found in a pure species population, which may contribute to their elevated community effects. I also showed, through common garden surveys in multiple years, that genetic variation in a single species is capable of strongly influencing the organization of arthropod community assembly. From this same population I

discovered multiple underlying genes and their potential biological functions contributing to

several different levels of community organization. Determining if functionally similar genes

have the same outcome in other pure species of Populus will enhance the predictive power of

host plant genetics on ecosystem assembly. Finally, I found that community assembly of insect

floral visitors and herbivores is detectable in a natural population of S. nigra and is mediated by a

multitude of genetically controlled characteristics, including dioecy, volatile organic compounds,

and catkin/leaf metabolites. Additionally, I found abnormal seasonal temperature created an

asynchrony in catkin bloom and insect emergence, which has the potential to alter tree-pollinator relationships in early spring. Overall: plant genotypes dictate plant physiology and morphology in ways that structure the greater biotic community. Host plant genetic effects combine with spatial and temporal environmental variability to drive dynamic plant-pathogen-herbivore- pollinator interactions.

166

5.6 Literature cited

Abdullah AS, Moffat CS, Lopez-Ruiz FJ, Gibberd MR, Hamblin J, Zerihun A. 2017. Host–

multi-pathogen warfare: pathogen interactions in co-infected plants. Frontiers in Plant

Science 8: 1806.

Bartomeus I, Park MG, Gibbs J, Danforth BN, Lakso AN, Winfree R. 2013. Biodiversity

ensures plant–pollinator phenological synchrony against climate change. Ecology Letters

16: 1331–1338.

Koskella B, Giraud T, Hood ME. 2006. Pathogen relatedness affects the prevalence of within-

host competition. The American Naturalist 168: 121–126.

Muchero W, Sondreli KL, Chen JG, Urbanowicz BR, Zhang J, Singan V, Yang Y,

Brueggeman RS, Franco-Coronado J, Abraham N, et al. 2018. Association mapping,

transcriptomics, and transient expression identify candidate genes mediating plant-

pathogen interactions in a tree. Proceedings of the National Academy of Sciences of the

United States of America 115: 11573–11578.

Orton ES, Brown JKM. 2016. Reduction of growth and reproduction of the biotrophic fungus

Blumeria graminis in the presence of a necrotrophic pathogen. Frontiers in Plant Science

7: 742.

Tschaplinski TJ, Standaert RF, Engle NL, Martin MZ, Sangha AK, Parks JM, Smith JC,

Samuel R, Jiang N, Pu Y, et al. 2012. Down-regulation of the caffeic acid O-

methyltransferase gene in switchgrass reveals a novel monolignol analog. Biotechnology

for biofuels 5: 71.

167

Vining KJ, Pomraning KR, Wilhelm LJ, Priest HD, Pellegrini M, Mockler TC, Freitag M,

Strauss SH. 2012. Dynamic DNA cytosine methylation in the Populus trichocarpa

genome: tissue-level variation and relationship to gene expression. BMC Genomics 13:

27.

Willmer P. 2012. Ecology: pollinator–plant synchrony tested by climate change. Current

Biology 22: R131–R132.

168